Virtual Fireside Chat & AMA with Yoni Brander, Chief Scientist & Esther Derman, Sr. AI Researcher at Innodata

​Capabilities Are All You Need:

​What Matters Next in LLM and Agentic Development

​Join a virtual fireside chat + live AMA with a leading AI researcher (and brilliant friend) for a blunt look at where AI is truly advancing—and where the industry is fooling itself.

​The Core Problem - Why It Matters

• ​Current debates over LLM scaling and architecture focus on parameters and generalized knowledge
  

• ​Datasets, RLHF, and Evals focus on domains and hope for emergent capabilities.
  

• ​Even those working to train models of specific capabilities have not cracked the code of real gains in that area in RL

​Until now? The solution—and what you’ll unlock in this fireside chat.

​The Solution - What you’ll learn

• ​A capabilities‑first framework for thinking about AI scaling intelligence gains across domains (reasoning, instruction following, planning, and tool use).
  

• ​A concrete mechanism to train agentic models to be more resilient and, by extension, unlock bigger, more generalizable gains per dollar.
  

• ​An approach that examines training agentic models to be demonstrably better at important capabilities, even when faced with ambiguity, adversarial content, long context, multi-hop reasoning, and unseen challenges.
  

• ​Where current RL/eval datasets are broken, and how that shows up as weird model failure modes in real products.
  

• ​The market failures no one wants to talk about—and the frontier opportunities they’re creating in AI post‑training.

​Who should join

​AI researchers, product leaders, and leaders who want a sharper lens on what actually moves the needle in AI.

​About the Speakers

​Yonatan “Yoni” Brander is an AI researcher, VP, and Chief Scientist at Innodata. His research is focused on agentic evaluation, adversarial robustness, and the science of building resilient AI systems that don't just perform—they hold up under pressure. He leads Innodata's Agentic Evaluation and Safety Practice, where his team builds the infrastructure, data pipelines, and datasets that frontier labs use to evaluate and improve their models and agents.

Esther Derman is a Senior AI research scientist at Innodata. Esther has co‑authored multiple reinforcement learning papers with Shie Mannor (Distinguished Scientist, Nvidia), including work on Bayesian approaches to robust reinforcement learning and twice‑regularized Markov decision processes. https://scholar.google.com/citations?user=IBIXZCAAAAAJ&hl=fr

About the host:

​​​​​​​​​​​​​​👋 Ken Morimoto is a veteran startup & tech operator (ex-Amazon, Scale, NewEdge, Bigband, Cobalt), active angel/VC, community builder, Director at Innodata, Chair of AI Circle (Seattle), and GP at Leading Edge VC. He’s active in the Seattle, NYC, and Bay Area tech communities and has been actively hosting AI Leaders & Builders meetups since 2020.