Reimagining Proteins with AI: Antibodies and Beyond
AI has changed what’s easy in protein engineering. Structure prediction, de novo binder and antibody design, developability and immunogenicity triage, sequence optimization - work that took years now takes months and even days. The harder question is what that actually buys us. A pipeline moves at the speed of its slowest step, and a step getting cheaper only shortens the timeline if that step was the one holding things up.
Our discussion takes that question seriously across the modalities defining the field today - antibodies, ADCs, nanobodies - and look into the future of engineered proteins: transcription factors, enzymes and proteins with previously unseen properties. Walking from sequence to clinic, we look at each step and ask how AI already deliver on safer bets like antibodies and where there is an opportunity to shack up the field of protein design to accelerate research, answer biological questions and expand treatment options.
Meet the speakers:
Kavita Kulkarni is a Principal PM for AI Research at Biohub, where she works on biological foundation models. Previously at Google Research, she was Global Head for Scientific collaborations in Biomedical Intelligence and co-invented the AI Co-Scientist, in addition to contributing to MedPaLM, Med-Gemini, and AMIE. She holds Stanford graduate fellowships in BioDesign & Innovation and Biomedical Informatics, and her work sits at the intersection of AI, biology, and clinical impact.
Nikita Savelyev is MSAT scientist with 8 years of expertise in production of biologics, in particular ADCs. He brings up expertise in optimizing ADC production, in particular - payload and linker chemistry and bioconjugation.
RIco Meinl is product lead for applied AI team at Retro Biosciences. His team designed new versions of Yamanaka factors to convert somatic cells into stem cells more efficiently. He and his team now integrate AI and robotics to engineer transcription factors with the goal to develop therapies that extend human health and lifespan.
James Lucas is Associate Director of Computational Protein Generation at Generate:Biomedicines, where he leads de novo protein design strategy and a team developing scalable computational approaches for therapeutic discovery. His work connects generative models with computational evaluation and high-throughput experimental testing, enabling rapid design-build-test-learn cycles. Before joining Generate, James was a computational protein design scientist at Nautilus Biotechnology. He holds a Ph.D. in Bioengineering from UC San Francisco and UC Berkeley. His interests include model-guided protein engineering and translating advances in machine learning into experimentally validated protein therapeutics.