Rahul Kapil

Seasoned AI & Data Science leader with 15+ years in the field. Last 2 years spent building end-user products on LLMs and unstructured data. Currently leading: AI org design and growth: hiring senior ICs and managers, defining career paths, and shaping the research / engineering / product — Product strategy for Agentic AI use cases — copilots, conversational interfaces, document intelligence, semantic search, agentic workflows — owned from problem framing through GA and post-launch iteration — Technical strategy: model selection, build-vs-buy, post-training investment, infra and orchestration choices, and the evaluation philosophy— Cross-functional partnership with product, design, GTM, legal, and the executive team — translating model capability into roadmap, and roadmap into measurable customer and business outcomes Currently Building: Retrieval systems over unstructured corpora (PDFs, transcripts, emails, code, multimodal): parsing, chunking, embeddings, hybrid search, reranking, grounding — Agentic systems with tool use, planning, memory, and the guardrails that make them safe in production — Post-training programs: SFT, preference tuning, distillation, and structured-output reliability for domain-specific behavior — Evaluation infrastructure: offline benchmarks, LLM-as-judge, regression suites, and online experimentation — the loop that turns demos into dependable products — Production inference at scale: latency, cost, observability, and failure-mode containment How I operate: Strong opinions on the unglamorous parts — data quality, eval rigor, error taxonomies, and the gap between a working demo and a dependable product — and I push the org to invest there. Keeping the loop tight between research, engineering, and customer-facing reality, and trust ICs with real scope. Equally comfortable in a code review, a roadmap debate, or a board u