newsfilter.io

Ion Stoica

Showing 1–5 of 5 transcripts.

  1. RAISE Summit41 min

    From Research to Reality: Why Open Source Is the Engine of AI | RAISE Summit 2026

    Ion Stoica, Thomas Wolf, Dillon Rolnick, Eiso Kant, Robin Rombach, Mark Porter, Dylan Rolnick

    Founders from Poolside, Databricks, Neuse Research, and Hugging Face, moderated by Mark Porter, convened to debate preventing AI monopoly through open-weight models rather than closed ecosystems. The panelists argued that open source remains the only viable path for critical sectors like robotics and scientific discovery while highlighting risks of geopolitical bias and regulatory lockout under current centralized control. Ultimately, the group emphasized a strategic shift toward commoditized infrastructure and real-world evaluation to avoid security fears and drive the next wave of physical and scientific breakthroughs.

  2. RAISE Summit21 min

    Opening Fireside: Ramin Hasani, Liquid AI & Ion Stoica, Arena | RAISE Summit 2026

    Ramin Hasani, Ion Stoica, Rachel Metz, Ayaan, Ian

    Led by MIT-derived algorithms and advised by Databricks co-founder Ian, Liquid AI is deploying energy-efficient, hardware-agnostic foundation models designed to run locally on consumer-grade chips rather than massive data centers. This approach addresses a shifting market where enterprises prioritize security and regulatory control by replacing expensive, closed-source systems with open-weight alternatives that offer near-parity in performance for specific tasks like coding and office work. The discussion further highlights a strategic pivot toward specialized "cities" of smaller, deterministic models that interact within agentic pipelines to ensure reliable, human-liability-backed deployment across industries.

  3. a16z1h 45m

    Beyond Leaderboards: LMArena’s Mission to Make AI Reliable

    Anjney Midha, Anastasios N. Angelopoulos, Wei-Lin Chiang, Ion Stoica

    LM Arena has transformed from a static benchmark into a dynamic "humanity's exam" that evaluates over 280 AI models through real-time feedback from one million monthly users, effectively eliminating data contamination through fresh prompt generation. By treating evaluation as Reinforcement Learning rather than Supervised Learning, the platform utilizes techniques like "style control" and the open-sourced "Prompt-to-Leaderboard" router to achieve twice the performance-per-cost while maintaining academic neutrality. Looking forward, the organization plans to expand into private industry-specific Arenas and multi-modal agent testing while remaining committed to open-sourcing all data and research to preserve ecosystem trust.

  4. Sequoia Capital1h 0m

    Turning Academic Open Source into Startup Success ft Databricks Founder Ion Stoica

    Ion Stoica, Stephanie Zhan, Sonya Huang, Jan Stojka, Matej

    Databricks addresses the critical gap between AI experimentation and production by promoting Compound AI Systems and launching the open-weight Dbricks model to ensure enterprise data privacy and compliance. Under founder Jan Stojka's guidance, the company leverages strategic partnerships with rivals like Microsoft while prioritizing robust data infrastructure over base model capabilities to drive accuracy and security. Looking ahead, the organization forecasts a shift toward commoditized inference costs and autonomous agents, urging founders to focus on verifiable, production-ready solutions rather than technical demonstrations.

  5. a16z10 min

    a16z Podcast | On Data and Data Scientists in the Age of AI

    Ion Stoica, Scott Clark, Frank Chen, Jan Stojka

    Enterprises successfully operationalize AI by progressing through a data foundation, operationalization, and integration phase while avoiding pitfalls like data integrity issues and contextual misalignment. Strategic success relies on an "all-in" approach that prioritizes portfolio hedging, shares artifacts for productivity, and leverages modern tooling to reduce time-to-market by an order of magnitude. As infrastructure barriers vanish, the data science role shifts from algorithm construction to defining business context, ensuring that commoditized tools optimize verified objectives rather than merely accelerating incorrect outcomes.