Smart Biotech Scientist | The CMC and Bioprocessing Podcast for Process Development and Manufacturing Leaders

Is your process model gathering dust while real-world decisions march on without it? Many technically strong digital tools—no matter how powerful—end up sidelined in bioprocess development. Why is that, and what does it actually take for digital modeling to earn its place at the lab bench?
In this episode, David Brühlmann sits down with Moritz von Stosch of HINA Bioventures, a scientist-turned-investor who’s navigated both the technology provider and pharma end-user sides. Moritz von Stosch brings firsthand insights from building digital solutions, deploying them in pharma, and evaluating what actually sticks when the clock is ticking and the stakes are high.
Topics discussed:
The importance of starting with your purpose when selecting a modeling strategy (03:00)
Experimental design: when to use factorial versus random space-filling designs, and the inherent biases each can introduce (04:22)
The future of agentic workflows, AI, and autonomous labs in bioprocessing, including the importance and challenges of sensor technologies (06:02)
Practical hurdles to implementing fully autonomous laboratories, and the non-obvious value humans still bring (09:09)
Adapting mindsets for the evolving role of the bioprocess engineer in a more automated, agent-driven workflow (11:48)
Advice for biotech professionals on how to assess the real value in new technologies versus hype, and the role of proprietary algorithms and company stability (13:33)
How cultural factors, not just technology, shape the adoption of modeling and digital tools in biotech (16:02)
Strategies to help the bioprocess community adopt more innovation while ensuring compliance in regulated environments (17:14)
Three key takeaways for staying relevant—including generating data for tomorrow’s questions and balancing curiosity, experimentation, and oversight (19:14)
Smart insight: Three takeaways from Moritz von Stosch:
Generate data for tomorrow’s questions, not just today’s.
Be curious—try, adapt, don’t over-plan.
Stay relevant: Use digital tools but keep your human edge (creativity, critical thinking).
To keep exploring how modeling and prior knowledge can speed up process development, check out these related conversations:
Episodes 05 - 06: Hybrid Modeling: The Key to Smarter Bioprocessing with Michael Sokolov
Episodes 173 - 174: Mastering Hybrid Model Digital Twins: From Lab Scale to Commercial Bioprocessing with Krist Gernaey
Episodes 277 - 278: Your Bioprocess Data Already Holds 35% More Yield: From End-to-End Models to Digital Twins with Ignasi Bofarull-Manzano
Episodes 215 - 216: From Data Silos to Autonomous Biomanufacturing: Digital Twins and AI-Driven Scale-Up with Ilya Burkov
Connect with Moritz von Stosch:
 Website: www.hinabioventures.com
LinkedIn: www.linkedin.com/in/movosto
Support the show

Are you frustrated with reinventing the wheel every time a new molecule enters bioprocess development? Is there a better way to harness data and modeling to accelerate breakthroughs in biotech? The bioprocessing industry is constantly searching for ways to streamline development, reduce costs, and speed new therapies to market—yet leveraging existing knowledge remains a significant challenge.
In this episode, David Brühlmann welcomes Moritz von Stosch, partner at HINA Bioventures and a pioneer in digital bioprocess development. Spanning roles from academic research in hybrid modeling to solution design at DataHow and now biotech investment, Moritz von Stosch’s journey traces how mechanistic, hybrid, and machine learning models have (and haven’t) changed process development over the last two decades.
Topics discussed:
Moritz von Stosch’s career: from early interest in modeling to investment in digital bioprocess tools (03:36)
The reality and limitations of platformization and knowledge reuse in large pharma (07:37)
Historical context and effectiveness of QbD (Quality by Design) and PAT (Process Analytical Technology) initiatives (10:24)
Where modeling can reliably deliver value today in process and clinical development—and where it still overpromises (12:17)
The importance of starting to build models early and collecting data that can be reused and repurposed across the process lifecycle (14:17)
Approaches to transferring learning between molecules, cell lines, platforms, and the fundamental limitations of this transfer (16:40)
Opportunities and difficulties in transferring process knowledge across scales (from microtiter plates to large bioreactors) (18:31)
Smart insight: David Brühlmann and Moritz von Stosch agree that models are only useful when they are built on reliable data and interpreted with expert judgment. Models can help biotech teams make better decisions, but they don’t replace human expertise. The real value comes from looking at the whole process, focusing on the data that matters, and making sure knowledge is shared effectively. This can help teams improve bioprocess development and bring new treatments to patients sooner.
To keep exploring how modeling and prior knowledge can speed up process development, check out these related conversations:
Episodes 05 - 06: Hybrid Modeling: The Key to Smarter Bioprocessing with Michael Sokolov
Episodes 173 - 174: Mastering Hybrid Model Digital Twins: From Lab Scale to Commercial Bioprocessing with Krist Gernaey
Episodes 277 - 278: Your Bioprocess Data Already Holds 35% More Yield: From End-to-End Models to Digital Twins with Ignasi Bofarull-Manzano
Episodes 215 - 216: From Data Silos to Autonomous Biomanufacturing: Digital Twins and AI-Driven Scale-Up with Ilya Burkov
Connect with Moritz von Stosch:
 Website: www.hinabioventures.com
LinkedIn: www.linkedin.com/in/movosto
Support the show

The entire public record of antibody affinity data comes to about half a million measurements, and by David Younger's estimate, upwards of 75% of it is AlphaSeq data from his own company's publications. His conclusion is that no single company can close the gap alone. So he is asking pharma and AI labs to pay for an experiment and share every result with the other members.
In Part 2 of this two-part conversation, David Brühlmann talks with Younger, founding CEO of A-Alpha Bio, whose AlphaSeq platform uses engineered yeast to measure protein interactions at scale. Part 1 set out the shortfall: about 10,000 antibody-antigen structures in the Protein Data Bank against a need of hundreds of thousands to millions. Here Younger explains how the consortium keeps proprietary targets out of the shared pool, who has joined, and why he believes it differs from federated learning.
Topics discussed:
Challenges faced by individual companies in generating sufficient data for effective AI models (02:17)
How a data consortium enables pre-competitive data sharing while protecting proprietary information (06:13)
Profiles of ideal consortium members and examples of current participants from pharma and AI model builders (08:36)
The consortium’s current focus on VHH antibodies and plans for expansion to broader datasets and applications by 2030 (10:00)
Comparison between the consortium approach and federated learning models in data sharing and collaboration (12:01)
The main takeaways about the quantity and quality of data needed for machine learning in protein engineering (14:09)
Smart insight: As David Younger envisions it, successful maturation of these consortia could fundamentally shift the field by 2030. Multiple consortia could address different scaffolds, data types, and problem statements, alleviating the bottleneck that has long constrained progress. The hope is that abundant, high-quality data will enable breakthrough therapeutic advancements while revealing new gaps that, in turn, inspire the next wave of innovation. Notably, David Younger sees collaboration—not just data scale—as the true secret weapon. Unlike federated learning, consortia incentivize members to share not just data, but experimental insights and best practices, effectively lifting all boats and accelerating collective progress.
Curious how others are tackling the data problem in protein design? These episodes pick up right where David Younger leaves off.
Episodes 185 - 186: How Generative AI Accelerates Protein Design for Biotech Discovery with Elise de Reus
Episodes 115 - 116: Revolutionizing Biologics Development with Hyper Throughput Screening and AI with Jeremy Agresti
Episodes 123 - 124: Manufacturability: Why Most Protein Candidates Fail (And How to Pick Winners Early) with Susan Sharfstein
Episodes 219 - 220: From 10,000 Structures to 1.8 Billion Interactions: Breaking the Data Bottleneck to Engineer Efficacious Therapeutics with Troy Lionberger
Connect with David Younger
 Company website: www.aalphabio.com
LinkedIn: www.linkedin.com/in/david-a-younger/
Data Atlas website: www.atlas.aalphabio.com
Support the show

The Protein Data Bank holds about 10,000 antibody-antigen structures and grows by roughly 1,000 a year. David Younger's scaling work suggests that generalizable de novo antibody design needs hundreds of thousands to millions. At that pace, even 200,000 structures is nearly two centuries away.
In this episode, David Brühlmann hosts David Younger, founding CEO of A-Alpha Bio and the mind behind the AlphaSeq platform. Trained at the intersection of synthetic biology and computational protein engineering, David Younger cut his teeth at the Institute for Protein Design and sharpened his focus on the stubborn gap between theoretical promise and practical performance. His approach rewrites the rules of data generation—scaling up both the volume and the insight from protein interaction data to deliver the kind of information AI needs to actually make an impact in therapeutic design.
Topics discussed:
Bottlenecks caused by insufficient and low-quality data in AI-enabled protein engineering (02:53)
The limitations of traditional protein discovery methods and the persisting role of serendipity in antibody development (08:02)
An overview of the AlphaSeq platform and how it enables large-scale measurement of protein interactions using engineered yeast (10:50)
The differences between measuring protein-protein affinities with conventional techniques and AlphaSeq’s batch approach, including cost and speed advantages (16:54)
Strategies for generating training data for machine learning models in protein engineering, including the need for structural and affinity data (18:23)
The value and limitations of the Protein Data Bank (PDB) in training de novo design models (20:17)
The challenge of benchmarking foundational models for protein engineering and the need for universal industry standards (23:11)
Smart insight: The future of biotherapeutic innovation is intertwined with a simple truth: more and better data unlock AI’s true potential. With breakthrough wet-lab technologies like AlphaSeq, and commitment to openness and benchmarking, the dream of efficient, precise, and truly rational protein engineering may soon become reality. The next big leap in biotech is not just smarter algorithms—but the scale, quality, and transparency of the data that powers them.
Curious how others are tackling the data problem in protein design? These episodes pick up right where David Younger leaves off.
Episodes 185 - 186: How Generative AI Accelerates Protein Design for Biotech Discovery with Elise de Reus
Episodes 115 - 116: Revolutionizing Biologics Development with Hyper Throughput Screening and AI with Jeremy Agresti
Episodes 123 - 124: Manufacturability: Why Most Protein Candidates Fail (And How to Pick Winners Early) with Susan Sharfstein
Episodes 219 - 220: From 10,000 Structures to 1.8 Billion Interactions: Breaking the Data Bottleneck to Engineer Efficacious Therapeutics with Troy Lionberger
Connect with David Younger
 Company website: www.aalphabio.com
LinkedIn: www.linkedin.com/in/david-a-younger/
Data Atlas website: www.atlas.aalphabio.com
Free 5-day email course, The CMC Failure Chain: the five recurring CMC mistakes that put your promising program at risk → Get it here
Support the show

Cardiovascular disease has long been managed by slowing its relentless progression. But Matthew O'Connor, CEO of Scientific Affairs at Cyclarity Therapeutics, is betting on an outright paradigm shift: treating atherosclerotic plaque with therapies designed to actively reverse disease, not simply manage symptoms or risk.
In this episode, David Brühlmann explores the strategy, regulatory journey, and manufacturing realities behind Cyclarity's unconventional cyclodextrin-based approach—a molecule sitting at the crossroads of small and large molecule drug development.
Topics discussed:
Insights into the company's clinical development progress, including completion of a first-in-human study and what comes next (02:47)
How the FDA views drugs that don't fit traditional categories and how this influences development pathways (04:35)
The challenges and considerations in demonstrating efficacy for drugs designed to reduce heart attacks and strokes (05:22)
Cost considerations in manufacturing the cyclodextrin dimer and how it compares to biologics and GLP-1 drugs (08:14)
Vision for how cardiovascular disease may be diagnosed and treated a decade from now, including the role of imaging and prevention (10:24)
Broader implications for treating age-related diseases by targeting toxic metabolites in the body (14:25)
Why next-generation therapeutics may be built not by sticking to old modalities but by tackling root causes with custom solutions (16:10)
Reflections on making new drugs both safe and fundamentally beneficial, minimizing side effects (17:10)
Smart insight: Start with the problem, not the modality. Matthew O'Connor argues that instead of defaulting to an antibody, a small molecule or a gene therapy, developers should build a custom solution to the problem they're trying to solve. A drug that suppresses a pathway across the whole body brings side effects you have to manage. A drug that removes something harmful, like oxidized cholesterol, can have side effects that are actually beneficial. His goal for the next generation of drugs is ones that simply make people healthier.
When a drug is too big to be a small molecule and too small to be a biologic, every CMC decision gets harder. These episodes look at what it takes to develop, characterize and manufacture a modality that doesn't fit an existing box.
Episodes 281 - 282: When Your Delivery Vehicle Contains a Membrane Protein: CMC Decisions With No Regulatory Precedent with Jitendra Kumar
Episodes 207 - 208: Cryogenic Infrared Ion Spectroscopy: From Mass Spec Limitations to Molecular Precision with Tom Rizzo
Episodes 197 - 198: Mastering ADC Development: CDMO Strategies for Analytics and Scale-Up with Amanda Hoertz
Episodes 231 - 232: From IND to BLA: The Biologics CMC Decisions That Determine Regulatory Success with Henri Kornmann
Connect with Matthew O'Connor:
Website: www.cyclaritytx.com
LinkedIn: www.linkedin.com/in/droki
Email: matthew.oconnor@cyclaritytx.com
Free 5-day email course, The CMC Failure Chain: the five recurring CMC mistakes that put your promising program at risk → Get it here
Support the show