At Kenoma.ai, I co-authored an LLM platform that turns company communications and documents into answers with attached evidence. I built its grounding, ingestion, and evaluation layers as the platform moved from an R&D proof of concept to production.
Since August 2026, I’ve run an independent R&D line at Kenoma.ai focused on small task-specific models and a local, single-binary build. I trained its claim-extraction and session-triage models, and built the reference set and match-and-coverage evaluation used to assess revisions.
Before Kenoma.ai, I worked on applied AI R&D at Involve.software, including clinical workflows with live transcription and doctor-in-the-loop document filling. I’ve also built backend systems for fintech infrastructure and anti-fraud scoring, and published a reproducibility package for research on public feeds.
Argentina