The Mack Institute’s Commercialization of Academic Science (MGMT 8920-002) is hands-on course where MBA student teams collaborate with Penn researchers to explore the real-world potential of early-stage technologies. Students gain practical experience applying commercialization frameworks at the intersection of strategy, innovation, corporate venturing, and entrepreneurship.
Course Description and Guidelines
This 1.0-credit experiential learning course matches small teams of MBA students with innovative technologies developed by Penn researchers. Students will engage directly with inventors to understand the science behind early-stage innovations and explore their potential for commercialization. In some cases, established corporations may also be involved, providing additional industry context and pathways to impact. Through a combination of instructor-led discussions on commercialization frameworks and independent team research, students will gain hands-on experience at the intersection of strategy, innovation, corporate venturing, and entrepreneurship.
Students must apply to be considered for the course. Applications are due August 26, 2026. The course is open to second-semester and second-year MBA students; Wharton Executive MBA (WEMBA) students may also be considered. Ideal candidates have experience or a strong interest in commercialization, innovation, or entrepreneurship.
Fall 2026 Projects
To read full project descriptions, click the titles.
1) Vividence Labs: Building and Validating the Customer Business Case
Growers make irrigation, crop protection, and field-management decisions with incomplete information. Existing tools can show that soil or weather conditions are changing, but they do not always reveal what the plant itself is experiencing early enough to prevent a loss.
Vividence Labs is developing a low-cost, biodegradable, battery-free sensor that attaches directly to a leaf and can be read remotely. The promise is straightforward: detect plant stress earlier, help growers act sooner, and improve the economics of crop management.
A potential beachhead application is precision irrigation in citrus orchards. Precise irrigation is becoming increasingly important as water supplies grow more constrained and water stress threatens fruit size and yield. These effects can be especially severe in disease-affected orchards, including those affected by citrus blight.
Today, growers typically manage automated drip irrigation using weather-station data and a limited number of root-zone soil-moisture sensors. These tools provide useful information about environmental and soil conditions, but they do not directly measure how individual plants are responding. Because the sensors are relatively sparse, they can also miss meaningful variation across an orchard.
Vividence’s leaf sensors, developed at Penn through the IoT4Ag NSF Engineering Research Center, could add an earlier and more plant-aware signal of drought stress. Their low cost could allow them to be deployed at higher density, giving growers a finer-resolution view of plant conditions and potentially enabling more precise irrigation decisions.
The opportunity is compelling, but the customer economics have not yet been fully established. Vividence needs to compare how a citrus grower manages irrigation today with what would change after adopting plant-level sensing. The analysis must capture the full cost of deploying the technology, the decisions it could improve, and whether the resulting value is large enough to justify adoption. Building and validating that business case is the project.
The question. What is the business case for using Vividence to support precision irrigation in citrus orchards, and does that case hold when tested against the operating and economic reality of growers? By the end of the semester, the founder should have a clear model built from customer interviews and other credible inputs, with the most consequential assumptions tested.
Where you would start. The business case needs to answer four fundamental questions.
- How do citrus growers monitor plant water needs and manage irrigation today, and where does the current approach fall short?
- What would it cost to deploy and operate Vividence, including sensors, installation, readout infrastructure, software, maintenance, and labor?
- What irrigation decision would earlier, plant-aware information change, and what measurable outcome could improve, such as water use, fruit size, yield, crop quality, or avoided plant loss?
- What level of evidence, economic value, and payback would make a grower willing to test and ultimately adopt the technology?
The work. You would first build a current-state versus Vividence-enabled economic model for an agreed citrus-orchard use case. Initial inputs would come from the company, available crop and irrigation data, and expert estimates.
You would then conduct 15 to 20 interviews with citrus growers, farm managers, irrigation managers, agronomists, extension specialists, and other relevant experts. Early interviews would establish how irrigation decisions are made, what growers currently spend, what risks matter most, and where better plant-level information could create value. Later interviews would test whether the model’s assumptions, logic, and projected results reflect how customers actually evaluate new agricultural technologies.
The team would revise the model as the evidence improves, identify the assumptions that have the greatest effect on customer value, and recommend a focused pilot for validating them. That pilot might test factors such as the lead time provided by the sensor, the number of sensors required per acre, changes in irrigation decisions, water savings, crop outcomes, and the total cost of deployment.
You would hand over a clear Excel business case, an assumption and source log, a summary of what the interviews confirmed or challenged, and a short plan for the next round of customer and field validation.
How you would work. The team would meet weekly with founder Dr. Akhila Mallavarapu. The founder and technical team would provide product, manufacturing, and deployment estimates. Customer and expert interviews would provide the operating and economic evidence. Every important input should have a source, and every major claim should be tested.
Who we are looking for
- Someone who can conduct thoughtful customer interviews and move beyond general statements to uncover how farm decisions are actually made.
- Someone who can build a clear Excel model and explain the reasoning behind every important assumption.
- Someone comfortable learning enough about citrus production, precision irrigation, and farm economics to understand what would make the technology valuable and adoptable.
2) SMART HABIT: Prototyping a Next-Generation Self-Management Tool for Chronic Kidney Disease
People with kidney disease are asked to monitor their blood pressure and stay active, and most of them do not. A Penn team tested whether a smartphone program could help, and found that patients were willing to use one even when they had little experience with health technology. What they built was designed to answer a research question rather than to serve patients over the long run, and it moved behavior further than it moved health. The next challenge is to reimagine what a patient-centered, scalable digital platform could look like if it were designed to improve long-term self-management, support clinical decision-making, and create value for patients, providers, and health systems.
Background reading: Schrauben et al., Supporting Self-Management of Healthy Behaviors in Chronic Kidney Disease and Hypertension, CJASN 2024. https://doi.org/10.2215/CJN.0000000000000492
The question. What would a self-management tool look like if it were designed for patients and their clinicians from the start, and could it plausibly improve outcomes while creating value for patients, clinicians, and value-based healthcare systems rather than just enhance health behaviors? You would answer that by talking to patients and clinicians, and then by building something.
How you would work. You build the prototype yourselves using AI coding tools. There is no engineering team and no handoff. The point is to shorten the gap between what someone tells you in an interview and what they can try out a few days later. Prior coding experience helps but is not required. What matters is being willing to build something rough, watch it fail in front of a real patient, and rebuild it that week.
Roughly how the semester goes. Spend the first few weeks understanding the trial and interviewing the people it was built for, including their care partners and clinicians. Turn that into design directions and a first build. Then run weekly cycles of testing and rebuilding for as long as the calendar allows. Finish with a working prototype, the research behind it, and an honest view of what it would take to get this to patients.
A few constraints. Use synthetic data only. You are building a prototype rather than a medical device, so clinical claims are out of scope. Design for an older adult managing several conditions, since that is the actual user.
Who we are looking for
- Someone who builds with AI coding tools, or is eager to learn on a deadline.
- Someone who interviews well, which mostly means being patient and comfortable with silence.
- Someone with clinical or digital health exposure.
- Someone with design judgment who can tell a clean interface from a usable one.
- Someone who has shipped something before and knows what “good enough to test” means.
3) PROSPER: Scaling Pediatric Behavioral Health AI Without Losing Strategic Control
A team of researchers inside Children’s Hospital of Philadelphia is looking to commercialize an AI platform called PROSPER to get more kids into behavioral health care. The need is not in dispute. Demand has outrun supply for years, and families routinely wait months for appointments that should take weeks. The team has early funding, a small group of engineers, and a first product going live this winter.
Building the product is one problem. Scaling it is a different problem, and that is the one you would be working on.
PROSPER depends on things the hospital does not have and will not build. Cloud infrastructure and model hosting. Privacy tooling built for pediatric health data. Integration with the systems clinicians already use. Some route to provider organizations outside Philadelphia. All of it comes from outside companies, and most of those companies could build a version of PROSPER themselves if they decided the market was worth entering.
Distribution is where this gets uncomfortable. A health IT vendor that agrees to carry PROSPER also owns the customer relationship. That vendor can bundle a competing tool a year later, and PROSPER becomes one more feature inside someone else’s platform. Every partnership that speeds the platform up concedes something. Your job is to work out which concessions are worth making.
The question. How should PROSPER use technology partnerships to scale quickly while protecting what makes it valuable over the long run? By the end of the semester, leadership should know which capabilities to own and which to source, which organizations to approach and in what order, what each partner contributes and what each could take, and what the hospital should refuse to give up no matter how much speed it buys.
Where you would start. The team holds four working beliefs. Break the ones that deserve breaking.
- Compute and model infrastructure are close to commodities, and the risk is manageable through terms on training rights, data use, residency, and portability.
- Distribution offers the most leverage and creates the most exposure, so deal structure matters more there than anywhere else.
- The model is not what makes PROSPER defensible. The clinical expertise, pediatric datasets, and ability to generate evidence inside a major academic medical center are harder to copy.
- Several bounded partnerships would leave more room to maneuver than one large anchor relationship, even if the anchor moves faster.
The work. Map the market and show where customer access, technical dependency, and competitive overlap actually sit. Score candidates on what each adds and what each could take, expecting the most important names to rank high on both. Work out which of the hospital’s assets a partner genuinely could not reproduce, separating the ones that merely look proprietary from the ones that are. Recommend a structure for each priority relationship and name the protections it needs. Then turn all of it into something the founders can carry into a meeting.
You would hand over an ecosystem map, a scorecard covering named organizations, a defensibility analysis, a recommended partnership portfolio with sequencing and guardrails, and a partner-facing deck built for real meetings rather than for a grade.
How you would work.
Weekly sessions with the clinical founder and the technical lead. What you produce feeds partner and executive conversations happening this fall as the first product moves toward launch. A recommendation that makes commercial sense still has to clear patient protection, research governance, procurement, and a hospital’s reasonable caution about its own reputation. That is the condition under which anything gets built in an academic medical center, and learning to work inside it is most of the value here.
Who we are looking for
- Someone who knows healthcare or digital health, including the institutional realities that shape technology adoption in a hospital.
- Someone fluent enough in cloud architecture, machine learning, or data governance to judge technical fit and hold a credible conversation with an engineering lead.
- Someone from partnerships or corporate strategy who has watched a prospective partner decide from the inside.
- Someone who can impose order on a fragmented landscape and turn incomplete evidence into a call they will defend.
- Someone with exposure to venture formation, technology transfer, licensing, or early-stage deal design.
4) Hearth Therapeutics: What Would It Take to Raise the Next Round?
After a heart attack, reopening the blocked artery does not stop the scarring that can later lead to heart failure. Hearth Therapeutics is developing a targeted immunotherapy intended to reach injured heart tissue and help local immune cells clear damaged cells before fibrosis takes hold. The company has encouraging mouse data and is now moving toward larger-animal work.
The science gives investors something to look at, but the financing case is still taking shape. The larger-animal study is not complete. The company still needs the relevant cardiovascular rights to part of the technology, along with a credible path through toxicology, manufacturing, and clinical development. It is also not yet clear what the next round should pay for or which kind of investor is best suited to lead it.
That is the work here. The team is not being asked to collect compliments on the pitch deck. Investors often like a scientific story without deciding to fund it. The project is about finding the point at which interest becomes action, and understanding what is missing when it does not.
The question. What would have to be true for a serious investor to fund Hearth? By the end of the semester, the founder should know which investors fit the company, what evidence they would need, which gaps can be carried into a financing, and whether Hearth should raise now or strengthen the package first.
Where you would start. The company is making several assumptions that have not yet been tested outside the current team.
- The mouse results are strong enough to start serious investor conversations before the larger-animal study is complete.
- The proposed treatment window fits the way heart-attack patients are actually managed, with a clear physician and care setting responsible for treatment.
- There is a well-defined group of patients for whom the likely benefit and risk would justify a biologic.
- The company can secure the rights and freedom to operate needed for an institutional financing.
- The current team is sufficient for the next stage, or the missing development and operating experience can be added after financing.
The work. You would speak with investors who have backed preclinical therapeutics, cardiovascular companies, and immune-modulating drugs, along with corporate venture and business development teams. You would also interview interventional cardiologists, heart-failure physicians, imaging specialists, and clinical trialists. The founder and advisors would help open those doors.
With investors, the aim is to understand where the current package falls short and what would change the decision. What earns a meeting? What triggers real diligence? Is the larger-animal study essential, or merely helpful? Does the field license need to be complete? How much confidence is expected around manufacturing, regulatory strategy, and the management team? What should the next financing accomplish for the following round to make sense? The team should compare what investors say with the kinds of companies they have actually funded.
Rather than begin with a list of venture firms, the team would define the types of investors most likely to fit Hearth and then build the list from evidence. That may include specialist life-sciences funds, investors with cardiovascular or immunology experience, strategic investors, and groups comfortable with university spinouts and preclinical biologics. The point is not to create a long directory. It is to identify who could realistically lead, who might follow, and when each would be worth approaching.
Conversations with doctors should test whether the product has a plausible place in care, not whether the mechanism sounds interesting. Which patients would be considered? Who would make the treatment decision? Could the therapy be given within the proposed window, and in what setting? What outcome would matter enough to change practice? Which early trial measures would physicians trust? What safety concern would make them hesitate? The views of interventional cardiologists, heart-failure specialists, imaging experts, and trialists may not be the same, and the differences matter.
You would hand over a focused investor map, an account of the evidence each investor type would need, a clinician memo on patient selection and adoption, and a recommendation on whether Hearth should raise now or wait. The final work should also sharpen the financing story: what the next round buys, which risks it retires, and what the company must have in place before it begins.
How you would work. Regular sessions with Dr. Noor Momin, with Jon Epstein and other advisors helping with introductions and technical questions. The team would need access to the underlying preclinical package, current licensing status, development budget, and roadmap assumptions. This is research, not fundraising, and students would not solicit capital.
Who we are looking for
- Someone with enough life-sciences exposure to discuss a preclinical biologic without overstating what the data show.
- Someone who can interview investors and physicians and hear the difference between genuine interest and polite encouragement.
- Someone who can connect the science, clinical workflow, intellectual property, development plan, team, and financing.
- Someone willing to make a recommendation even when the evidence is mixed.
5) DTC-Flow: Platform for detection and isolation of disseminated tumor cells
After treatment for early-stage breast cancer, scans may show no remaining disease. Even so, a very small number of cancer cells can survive and hide in the bone marrow. These cells, known as disseminated tumor cells or DTCs, may remain inactive for years before causing the cancer to return. Doctors do not currently look for them as part of routine follow-up.
Penn researchers have developed DTC-Flow, a laboratory method designed to find these hidden cells in a bone marrow sample. In laboratory testing, it could detect one cancer cell among a million bone marrow cells. In an initial study of 42 high-risk breast cancer survivors, it classified 64 percent of patients as DTC-positive, compared with 29 percent using the older slide-based method. DTC-Flow can also separate the cells so researchers can study why they survive and which treatments might eliminate them. The results are promising, but longer follow-up and larger studies are still needed to show how well the test predicts recurrence or improves care.
A separate Penn clinical trial, CLEVER, used the older detection method and found that several existing drugs sharply reduced DTC counts in many participants. The trial did not test DTC-Flow, but it strengthens the case for a better way to identify patients with hidden disease and measure whether treatment is working.
This creates several possible businesses. DTC-Flow could first be used by pharmaceutical companies in clinical trials, by a specialized laboratory as a future recurrence-risk or treatment-monitoring test, or by researchers searching for new drug targets. It might later expand into other cancers or support the development of therapies aimed directly at dormant cancer cells. The challenge is that the test requires a bone marrow sample, specialized equipment, and more clinical proof before it can become part of standard care. The first market and regulatory path will therefore shape the larger opportunity.
The question
Which first customer, use case, and path to market give DTC-Flow the best chance of adoption while keeping the larger opportunity open? By semester’s end, the team should recommend one route and show what would need to be true for it to work.
Where you would start
- First customer and decision.Which group has the strongest immediate need: drug companies testing treatments, cancer centers following high-risk survivors, or researchers looking for new drug targets? What decision would DTC-Flow change, and who would pay?
- Comparable tools.Study other tests that detect tiny amounts of cancer, including blood-based tests and tests used after treatment in blood cancers. How did they prove value, gain regulatory acceptance, win payment, and reach routine use?
- Regulatory path.Could DTC-Flow begin as a research service or clinical-trial tool, then move into a specialized laboratory test and eventually an FDA-reviewed diagnostic? Which sequence creates value fastest without limiting future uses?
- Economics and value.Estimate the cost of collecting and processing each sample, the likely price, how often the test might be used, the size of the first market, and who would pay.
The work
The team would compare three commercial paths: a service for pharmaceutical trials, a clinical testing service, and a platform or licensing model that also supports drug discovery. Each path would be scored on customer need, access to bone marrow samples, evidence and regulatory requirements, willingness to pay, capital needs, ability to grow, and competitive advantage.
The team would conduct 12 to 15 interviews with oncologists, laboratory leaders, pharmaceutical trial teams, diagnostic and instrument companies, regulatory and payment experts, investors, and business-development leaders. Early interviews would test the problem, buyer, and workflow. Later interviews would test pricing, evidence, partnerships, investment interest, and exit routes.
The project would also consider expansion into other cancers, the discovery of new drug targets, and therapies aimed at dormant cancer cells. The semester ends with a commercial recommendation, not a product build, clinical study, regulatory filing, or drug-development program.
What you would deliver
- A recommendation for the first customer, use case, and business model.
- A practical roadmap for the evidence, regulatory steps, and expansion required.
- A financial model covering market size, cost per test, pricing, capital needs, and growth.
- A prioritized map of potential partners, investors, and exit options, including what must be owned, licensed, or supplied by partners.
How you would work
Weekly sessions with the Penn research and clinical team would provide the scientific evidence, current workflow, development history, and access to experts. Students would add the customer, competitive, regulatory, and financial evidence. The final recommendation must separate what DTC-Flow has shown, what CLEVER showed using the older test, and what a commercial product still needs to prove.
Who we are looking for
- Students interested in biotech, diagnostics, healthcare strategy, or venture creation. No oncology background is required.
- Strong interviewers who can explain complex ideas clearly and turn evidence into decisions.
- People comfortable with market research and basic financial modeling, and willing to compare several business paths.
6) CellFoundry Bio: Choosing a Path to Market for a New Way to Manufacture CAR-T
CAR-T therapy re-engineers a patient’s own T cells to recognize and attack their cancer, and for several blood cancers the results have been striking. Getting it to more patients has been the hard part. All-in cost runs about $1.5M per patient, roughly a quarter of manufacturing runs fail outright, and the wait between blood draw and infusion stretches four to six weeks, which is part of why only about one in five medically eligible patients ever receives the therapy.
CellFoundryBio, a Penn spinout, is working on the manufacturing end of that problem with a silicon microfluidic chip that delivers the genetic payload into T cells directly, removing the viral vector responsible for roughly 40% of manufacturing cost and a good share of the failures. The prototype data is published and the first silicon version is now in development.
What the company has not worked out is how any of this reaches a market. There are three plausible first customers, three plausible lead indications, and three plausible ways to structure a deal, and the market model that would help choose among them still rests on estimates rather than on conversations with anyone who might actually buy. Sorting that out is the project.
The question. Who is the first customer and under what commercial model? By the end of the semester the founders should be able to answer each of those with evidence behind it.
Where you would start. The company is operating on four assumptions, none of which has really been tested. Working out which ones hold is a good part of the job.
- The first customer is a contract manufacturer, on the theory that CDMOs carry vector cost directly and can adopt new hardware without running a clinical trial, though hospital manufacturing cores and pharma licensees are both live alternatives.
- Autologous CAR-T for large B-cell lymphoma is the indication as currently scoped, although allogeneic, NK, iPSC, or use as an LNP screening platform might offer a faster way in.
- A razor-blade consumable model captures the most value, priced against the roughly $270K per patient the chip actually addresses.
- The competitive field extends well beyond the other non-viral delivery startups, taking in electroporation incumbents like MaxCyte along with the general reluctance of any manufacturer to revalidate a process that already works.
The work. You would run 15 to 20 research conversations across CDMOs, hospital manufacturing cores, pharma process development groups, and payers, with the founder and scientific advisors opening most of those doors. What you learn there replaces the estimates in the current market model with figures that hold up. From there the work turns to mapping the competitive landscape more broadly than the obvious names, understanding what each type of customer values and would pay for, and identifying what would make adoption difficult for them. The recommendation at the end covers a first customer, a lead indication, and a commercial structure, checked against the terms of the Penn license. If there is time, you might also look at whether AI-driven quality control adds anything real to the platform.
You would hand over a market model built on primary research, a competitive landscape, an assessment of value and pricing by customer type, a recommended entry strategy with sequencing, and a revised commercial narrative the founders can use with partners and investors.
How you would work. Weekly sessions with the founder, who also leads IP strategy and partnership development. The scientific advisory board includes the fab that manufactures the chips, a Penn Medicine oncologist overseeing the validation work, and a GMP manufacturing expert. The company is early enough that a well-argued recommendation can genuinely change the direction it takes.
Who we are looking for
- Someone with life sciences or biotech exposure who is comfortable discussing GMP manufacturing and cell therapy workflows.
- Someone who understands platform, consumable, or capital equipment business models and how pricing works in each.
- Someone who conducts a good research interview, meaning they listen well and follow the interesting thread rather than the script.
- Someone who can build a market model from primary evidence and explain every assumption in it.
7) Conversational Biosignal Platform: Building a Defensible Beachhead
Connected medical devices collect rich physiological data, but most communicate little of it to the patient. A device may detect something important without being able to explain it, ask what happened, learn from the answer, or show how behavior affects health. Technology that enables patients and their wearable or implanted devices to communicate openly and bidirectionally could turn one-way monitoring into a continuous learning loop.
Penn researchers have developed and tested a platform in patients with epilepsy that combines EEG and clinical data with secure analytics and a conversational AI agent delivered through a smartphone app. Patients receive alerts, ask questions, and label events. In a pseudo-prospective analysis, those labels improved seizure detection and sharply reduced false alarms. The system also tracked signals associated with medication levels, cognition, sleep, and other activity important to patients. When uncertain, it can route data to an expert for adjudication and learn from the interaction.
The larger commercial vision is a governed intelligence layer that medical-device companies can embed to translate biosignals into insight and create a learning loop among device, patient, clinician, and algorithm. The same approach could differentiate neurostimulators, pacemakers, defibrillators, closed-loop insulin pumps, and new connected devices. A major study underway is testing whether measured data and patient dialogue can support therapy titration and remote programming of brain stimulators for Parkinson’s disease, reducing the need for patients to travel to specialty centers.
New LLM tools make some visible components, including chat, alerts, dashboards, and cloud orchestration, faster to build. The clinical system beneath them is not. Its safety and security architecture, biosignal analytics, expert-adjudication workflow, and execution know-how reflect six years of development. The core commercialization question is whether that accumulated capability can support a stand-alone platform company serving multiple device makers and hardware categories, or whether value creation and defensibility require a deep tie to a particular device, manufacturer, or condition.
The question. Which initial customer, use case, and product boundary offer the strongest route to market and the best opportunity to build a defensible moat? By semester’s end, the team should recommend one beachhead, determine whether the initial offering should be cross-device or device-specific, explain why that path wins, and identify which capabilities and rights must be owned, licensed, or secured through partners.
Where you would start. The commercialization case needs answers to four questions.
- Which customer, buyer, and workflow have the most urgent need for two-way interaction and adaptivebiosignal analytics?
- Should the company begin as a device-agnostic intelligence layer or a tightly integrated solution for one hardware platform or disease area? What are the technical, regulatory, and commercial trade-offs?
- What measurable value could the platform create, who captures it, and what proof would justify adoption?
- What could a capable device company reproduce within 12 months, whatremains difficult because of the team’s accumulated clinical and technical know-how, and how can data rights, validated models, safety and security, evidence, integrations, workflow, intellectual property, and partnerships compound into a moat?
The work. You would begin by mapping plausible device categories and commercialization models, then select three beachhead hypotheses for deeper testing. At least one should test a cross-device model and at least one a device- or condition-specific integration. The team would compare them on customer urgency, economic value, access, integration burden, evidence and regulatory requirements, platform reuse, and moat-building potential. Candidate entry points may include epilepsy devices, ambulatory EEG, clinical research, neurostimulation, cardiac devices, or other categories supported by the team’s evidence and access.
The team would run 12 to 15 interviews with product and medical leaders at device companies, clinicians, researchers, and a small number of regulatory, security, or digital-health experts. Early interviews would test the problem, buyer, workflow, product boundary, and value. Later interviews would test data rights, evidence requirements, integration, business model, cross-device reuse, and partnership structure. The team would then recommend one narrow route to market and show how it could expand across devices over time. The project does not include product development, a clinical study, or a regulatory filing.
You would hand over four connected outputs: a scored beachhead and product-boundary recommendation; a customer, workflow, and basic economic model for the leading use case; a copyability assessment and defensibility roadmap that separates existing know-how from assets still to be built; and an interview-backed ownership, partnership, assumption, and source log.
How you would work. Weekly sessions with the Penn research, clinical, and technical team. The team would provide the architecture, evidence, context, six-year development history, and security and safety assumptions. Interviews and desk research would supply the customer and competitive evidence. Every important claim should have a source. The final recommendation should separate what the prototype has demonstrated, what current studies are testing, and what a commercial product still needs to prove. It should identify what can be reused across devices, what must be customized, and what must be owned versus secured through partners.
Who we are looking for
- Someone who can conduct thoughtful interviewsacross clinical, technical, and commercial stakeholders.
- Someone who can turn a complex platform into a specific customer problem, product boundary, and economic case.
- Someone comfortable comparing cross-device and device-specific strategies and testing how safety, security, data, evidence, integrations, intellectual property, and partnerships can become a defensible advantage.
8) Hydropore: Building a Defensible Beachhead in Stationary Backup Power
Hydropore is developing a zinc-based solid “metal fuel” that stores energy and releases hydrogen on demand when mixed with water. Unlike conventional hydrogen systems, it is designed to avoid compressors and high-pressure gas storage, while keeping the material stable at atmospheric temperature.
The company initially explored hydrogen-powered drones, where longer flight time could create value. Customer discovery showed that the compressors and fueling equipment required by small users make that market unattractive for now. Hydropore is therefore shifting toward stationary, long-duration backup power.
The leading hypotheses are data centers, microgrids, and nanogrids, where customers may value long backup duration, a smaller footprint, easier siting, and less dependence on stored flammable gas. The alternatives include batteries, diesel or natural-gas generators, and conventional hydrogen systems.
Hydropore’s starting moat includes a patented materials synthesis process, more than 20 years of expertise, and trade-secret scale-up methods. The commercial challenge is to turn those assets into a system that is meaningfully cheaper, safer, easier to integrate, or more valuable than current options.
The question. Which initial customer, use case, and product boundary offer the strongest route to market and the best chance to build a defensible company? By semester’s end, the team should recommend one beachhead, define what Hydropore should sell, and state the cost, performance, partnership, and ownership conditions required to win.
Where you would start. The commercialization case needs answers to four questions.
- Which buyer has the most urgent need for long-duration backup, limited space, and protection from costly outages?
- Should Hydropore sell the metal fuel, a hydrogen-production module, a complete energy-storage system, or an integrated backup-power solution? What should partners provide?
- Does the business case work against batteries, generators, and conventional hydrogen when equipment cost, efficiency, backup hours, footprint, maintenance, safety, useful life, recharging, and downtime are included?
- What could a capable competitor copy, what is protected by patents and scale-upknow-how, and what additional advantages must be built through performance data, manufacturing learning, integration, service, and partnerships?
The work. Map the stationary backup-power market, current solutions, buyers, decision processes, partners, and business models. Test three beachhead hypotheses: data centers, microgrids, and nanogrids. Compare them on customer urgency, willingness to pay, required backup duration, footprint, integration burden, safety and permitting, sales cycle, access, economics, repeatability, and moat-building potential.
Conduct 12 to 15 interviews with buyers, microgrid developers, backup-power integrators, hydrogen operators, potential partners, and selected safety, regulatory, and manufacturing experts. Use early interviews to test the problem and requirements; later interviews to test the offering, economics, procurement, integration, scale-up, and partner roles. Build a transparent business-case model and recommend one focused route to market. The project excludes product development, a field pilot, detailed engineering, and regulatory submissions.
You would hand over four connected outputs: a scored beachhead and product-boundary recommendation; a customer and basic economic model for the leading use case; a copyability assessment and defensibility roadmap; and an interview-backed log of assumptions, sources, ownership decisions, partner needs, and go/no-go conditions.
How you would work. Meet weekly with the Hydropore research and technical team. Use its laboratory results, system assumptions, patent boundaries, material requirements, scale-up history, and product concept alongside interviews and desk research. Separate what has been proven in the lab, what remains an engineering assumption, and what a commercial system must still demonstrate. Identify what Hydropore must own and what it can secure through partners.
Who we are looking for
- Someone who can interview energy, infrastructure, technical, and commercial stakeholders and turn evidence into a clear recommendation.
- Someone who can translate a complex materials technology into a specific customer problem, offering, and business case.
- Someone comfortable comparing markets, building a basic economic model, and testing how intellectual property, manufacturing know-how, evidence, integration, service, and partnerships create defensibility.
9) Silencios: Bringing an Alzheimer’s Drug to the Clinic Without Giving Away the Platform Too Early
A research team at the University of Pennsylvania has developed a new way to treat diseases such as Alzheimer’s and Parkinson’s. The technology, called Silencios, uses small pieces of RNA to reduce the production of harmful proteins in the brain. Its unusual feature is that one drug can be designed to reduce more than one protein at the same time.
The first commercial program, AMY-001, is aimed at people with early Alzheimer’s disease. Most current treatments focus mainly on amyloid, one of the proteins associated with the disease. AMY-001 is designed to reduce the production of both amyloid and tau, the two major forms of Alzheimer’s pathology. The research team has produced promising mouse data showing that its dual-target approach can reduce both. Getting the science to work is one problem. Building a company that can afford to prove it in people is a different problem, and that is the one you would be working on.
The current animal evidence was produced by delivering the treatment directly into the fluid surrounding the brain. That is useful for testing the science, but it is not the product Amyndra ultimately wants to sell. The commercial plan depends on attaching the drug to a “brain shuttle” so that it can cross into the brain after an injection under the skin. Amyndra does not own that delivery technology. It is considering licensing it from another biotechnology company.
Delivery is only one dependency. Amyndra will also need outside manufacturers, clinical trial operators, regulatory expertise, trial sites, and eventually a company capable of selling an Alzheimer’s drug around the world. The team’s current financial model estimates that taking AMY-001 through development could cost approximately $741 million, most of it for clinical trials.
A large pharmaceutical partner could provide much of what Amyndra lacks. It could finance trials, manage regulators, manufacture the drug, work with insurers, and reach physicians and patients. It could also demand a large share of the economics, control over the development plan, rights to future uses of the drug, or access to the wider Silencios platform.
Timing is where this gets uncomfortable.
Partnering early would reduce the amount of money Amyndra has to raise and transfer some of the development risk to a larger company. But an early partner would be negotiating before Amyndra has human evidence, when the company has the least bargaining power.
Waiting could allow Amyndra to negotiate from a stronger position. But every additional milestone requires more capital, more time, and more exposure to failure. The company could spend heavily proving the science only to discover that the delivery system, safety profile, or clinical results are not strong enough to support a deal.
Every additional piece of evidence may make the company more valuable. It may also be the piece of evidence that shows the product does not work.
The question. How should Amyndra sequence development, financing, and pharmaceutical partnerships so that it can reach a convincing human proof point without giving away the Silencios platform too early?
By the end of the semester, leadership should know how far to take AMY-001 independently, which milestones would materially improve its negotiating position, how much capital would be required to reach each one, which organizations to approach and in what order, what rights Amyndra can afford to trade, and what it should refuse to give up.
Where you would start. The team holds four working beliefs. Break the ones that deserve breaking.
- Alzheimer’s is the right first market because its size and unmet need justify the cost, time, and scientific risk of developing a drug for it.
- The brain shuttle should be included from the beginning because an injection under the skin would be a major advantage for patients and physicians.
- Amyndra will receive substantially better partnership terms if it waits until it has early human data.
- A partner can receive broad rights to AMY-001 without weakening the rest of the Silencios platform, provided that future drugs and diseases are clearly excluded from the agreement.
The work. Map the path from the current research program to the first evidence in human patients that would cause investors or pharmaceutical companies to value Amyndra differently. For every major milestone, show how long it would take, what it would cost, what question it would answer, what could still go wrong, and what financing or partnership options it would open.
Compare at least three credible paths.
- One path may bring in a pharmaceutical partner before human testing. Another may raise enough private capital to complete the work required to begin clinical trials and then seek a partner. A third may keep AMY-001 independent through an initial human study or an early effectiveness result.
- Do not assume that the most ambitious path creates the most value. A faster or narrower first program might prove that the Silencios technology works in people with less capital. Amyndra has also developed programs aimed at tau, Parkinson’s disease, APOE4-associated Alzheimer’s, and other neurodegenerative conditions. Test whether one of those programs could provide a cleaner or faster proof of the platform, even if its immediate market is smaller.
- Map the organizations Amyndra may need along the way. This should include delivery technology companies, specialist investors, clinical development partners, manufacturers, and pharmaceutical companies with experience in neuroscience. Score each named organization on what it contributes, when Amyndra would need it, and what it could take in return.
Pay particular attention to the brain shuttle. Determine whether Amyndra should license one delivery system now, test several alternatives, negotiate a limited research agreement first, or work with a larger partner that already controls delivery technology. The answer should address ownership of improvements, access to future versions, manufacturing supply, exclusivity, sublicensing, and what happens if the relationship fails.
For potential pharmaceutical deals, distinguish between rights to AMY-001 and rights to the wider platform. Consider whether a partner should receive one disease, one geographic market, one drug, or a broader set of rights. Amyndra’s own materials suggest that a future partner could contribute commercial infrastructure, international regulatory capabilities, and manufacturing, but that such a deal could also leave Amyndra with a minority interest in the program.
Then make a call. Recommend one primary path and one fallback path. State what Amyndra should do next, what it should wait to do, and what result would cause the company to change direction.
You would hand over a development and value-creation map, a financing model covering the major strategic paths, a scorecard of named investors and partners, a recommended sequence of financing and partnership conversations, deal guardrails showing which rights Amyndra should retain, and a partner-facing deck built for actual meetings rather than for a grade.
How you would work.
Weekly sessions with the scientific founder and the commercialization team. What you produce should feed decisions about the brain-shuttle license, the next preclinical studies, fundraising, and pharmaceutical outreach.
A recommendation that looks attractive in a financial model still has to work within drug safety requirements, clinical trial realities, Penn’s intellectual property and licensing process, and the time required to produce credible scientific evidence. Those constraints cannot be removed from the problem. Working out how to build a company inside them is most of the value here.
Who we are looking for
- Someone with experience in biotechnology, pharmaceuticals, healthcare, or life-sciences investing.
- Someone who can understand the basic science well enough to distinguish what has already been demonstrated from what remains an assumption.
- Someone from pharmaceutical business development, partnerships, venture capital, or corporate strategy who understands how potential partners evaluate early drug programs.
- Someone with exposure to intellectual property, technology transfer, licensing, or early-stage deal design.
- Someone who can connect scientific milestones, capital requirements, negotiating power, and ownership into a recommendation the founders can act on.
Note: Projects will move forward based on sufficient student interest and enrollment. Because some projects may not receive enough student sign-ups to proceed, we strongly encourage you to select several projects that interest you when submitting your application.