Computer Science and Engineering student in Chile (graduating June 2028), working as a Digital AI Intern at Marsh, the world's largest insurance broker. I support AI adoption and build process automation for teams across Latin America using Power Platform, while gaining firsthand experience with how a global, regulated enterprise adopts generative AI.

My goal is to become an AI engineer building reliable systems on top of foundation models, with a focus on RAG pipelines, agents, and LLM evaluation. My current experience is in regulated industries, where understanding how the business actually works is often as important as the technology itself.

chilean-insurance-rag — under active development, building in public

A question-answering system over Chilean insurance regulation (CMF norms), with mandatory citations and an evaluation pipeline as a first-class component: golden dataset, retrieval metrics, and calibrated LLM judges. Built deliberately without orchestration frameworks (no LangChain / LlamaIndex) to understand and own every layer: ingestion, hybrid retrieval (BM25 + dense), generation, evals, deployment. Follow the commit history to see it grow.

Next in the pipeline: an agent for insurance document processing workflows, designed from publicly documented industry processes, and a reusable LLM evaluation harness.

LLM system evaluation, retrieval quality measurement, and production Python, learning in thin layers tied to what I'm building each week, not theory upfront. Currently reading AI Engineering (Chip Huyen).

Python · FastAPI · sentence-transformers · SQLite/sqlite-vec · bm25s · pytest · Docker · GitHub Actions · Power Platform (professional work)

I'm always happy to connect with people building practical AI systems, whether in startups, enterprise, or research. LinkedIn