About

I’m Jack Monte, an AI engineer building production LLM systems for B2B SaaS and 3PL/logistics companies.

My focus areas are retrieval-augmented generation (RAG), document and invoice automation, and AI agents. The common thread across all of them is a bias toward systems that are measured, not just demoed. Every project I ship carries an auditable evaluation dataset and published benchmarks, because in production the difference between “looks impressive” and “works reliably” is the whole job.

How I work

Evals first. Before a pipeline changes, it gets a baseline. Quality metrics (faithfulness, relevance, retrieval precision) are tracked separately from reliability metrics (error rate), so a regression in one never hides behind the other.

Observability built in. Every query is traced end to end with Langfuse, with scores attached to traces. When something degrades, the answer to “what changed” is in the data, not in guesswork.

Production patterns from day one. Typed data boundaries, centralized API clients with configured retries, and explicit error hierarchies. The goal is code a client’s team can read, extend, and trust after the engagement ends.

Stack

Python, FastAPI, ChromaDB and pgvector, Anthropic API, Langfuse. Development with Claude Code and full CI (ruff, mypy) on every project.

Proof

My open-source work and benchmark write-ups live on GitHub. The Projects page covers each build in detail.

Contact

The fastest way to reach me is LinkedIn.