200+ talks across 20+ countries—translating production ML and agentic AI into clear, actionable frameworks.
API Days Paris (Paris, France) — Speaker · Designing APIs for Agentic Systems
Book a Briefing in Paris →
Watch Google I/O Keynote Clip ↗
From being introduced on the Google I/O main stage by Sundar Pichai to delivering technical sessions at Codemotion Milan, Open Source Summit Europe, AGNTCon + MCPCon, and API Days, every talk centers on real architectural trade-offs, production postmortems, and concrete mental models.
"When NOT to Build an Agent"
Keynotes & founder summits
Executive & developer discussions
An architectural rubric grounded in real latency and cost postmortems—separating workflows best served by deterministic state machines from genuine multi-agent autonomy.
Stages: Codemotion Milan (Oct 2026) · AGNTCon + MCPCon Europe, Amsterdam (Sep 2026) · Driving AI Keynote, Copenhagen (May 2026)
Human-oriented APIs fail silently under stochastic tool use. How to design semantic MCP contracts, idempotency boundaries, and per-turn token circuit breakers.
Stages: API Days Paris (Dec 2026) · API Days New York (May 2026)
Replacing subjective vibe-checks with automated evaluation harnesses: PAIR LLM Comparator, trajectory assertions, Pass@k reliability, PII redaction, and injection defenses.
Stages: Open Source Summit Europe, Prague (2026) · AI_dev Europe (2025) · Milan0day (2026)
Architecture of an AI Racing Coach tested on a live track—pairing sub-15ms on-device Gemini Nano cues and physics safety gates with Cloud Run lap reasoning.
Stages: Google Cloud Next '26 (Las Vegas) · Google Track Day (Thunderhill Raceway)
Field lessons from billion-request ranking pipelines across Uber, Faire, and Twitter/X—and dataset curation practices from Sculpting Data for ML.
Stages: PyCon US · LeadDev Live · MLconf · Data Innovation Summit Stockholm
Speaker · Designing APIs for Agentic Systems
Speaker · When NOT to Build an Agent: Lessons from Systems That Didn't Want Autonomy
Speaker · AI Showdown: Quantitative LLM Evaluation
Speaker · When NOT to Use an Agent
AI Partner, Ambassador & Invited Speaker
Panel Moderator & Speaker · Agentic Reliability & Evals (with Anthropic, ClickHouse, Temporal & Cribl)
Keynote Speaker · When NOT to Use an Agent
Speaker · Weights, Worms, and Wreckage: Attacking AI
Panelist · Real-World AI Use Cases
Speaker · Designing APIs for Agentic Systems
Lead Workshop Facilitator · Executive AI Agents & No-Code Multimodal Assistants
Panelist & Featured Developer · Edge/Agentic AI & AI Racing Coach
Fireside Guest · Beyond Hype: Evaluating AI Agents
Invited Panelist
Invited AI/ML GDE & gDAB Builder · Gemini & ADK Global Summit
Speaker · Deepfakes, Generative AI & the Battle for Truth
Speaker · Guarding the LLM Galaxy
Speaker & Program Committee · AI Showdown: LLM Comparator
Speaker & Program Committee · Guarding the LLM Galaxy
Panelist · Agentic AI: AI That Reasons, Acts, Adapts
Panelist · Google Cloud Champion & GDE
Panelist · Responsible AI in Production
AI for Good Judge & 20 Years of Google Open Source Feature
Featured Developer on Main Stage (Introduced by Sundar Pichai)
Panelist (Building a Cloud-Powered Future) & Speaker (Guarding the ML Galaxy)
Keynote, Technical Speaker, Track Host & Program Chair
Invited Panelist, Guest Lecturer & Research Speaker
Agentic AI architecture ('When NOT to Build an Agent'), Agent-Native APIs & Cost Guardrails, Quantitative LLM Evaluation & AI Security, Split-Brain Edge + Cloud Real-Time AI, and Data-Centric ML.
San Francisco Bay Area (SFO / SJC), available for in-person international keynotes, executive summits, and virtual events.