Work

Forward deployed engineer on an enterprise AI platform that works with any foundation model and deploys on cloud or on-prem. I build on client data, run live demos, and work hands-on inside client environments.

• •Owned half the pipeline on a multi-week sprint for a public sector client, building automated validation of monthly Excel submissions against complex business rules and layout conventions. It surfaced tens of thousands of errors on real data, including thousands the client had missed. Completing the sprint moved the contract from development into production.
• •Ran multiple live demos for large audiences of government employees. Worked through discovery with the client to pin down what they needed changed, then built it out remotely inside their own VM.
• •Built an internal usage analytics platform in my first two weeks, replacing a manual process. It charts usage by organization and environment over any time range and sends scheduled, client-ready reports by Slack or email. Used by execs and account teams.
• •Shipped obfuscated Docker containers (PyArmor) so a client could deploy our solutions in their own environment while protecting our IP. Wrote the docs and the setup, run, and stop scripts.

Python · Docker · PyArmor · GitLab · SendGrid · Slack API · Bash

Led all AI initiatives for a 300-person engineering and construction consulting firm. I built on-premise AI infrastructure using llama.cpp for local LLM deployment and shipped production RAG systems, managing the full lifecycle from discovery through adoption.

• •Architected multi-model AI infrastructure using llama.cpp, choosing different Llama variants by task: long-context summarization vs. retrieval-augmented Q&A
• •Delivered the first production AI system in 6 months, with 3 more in development
• •Led an 18-person AI pilot program that included C-suite executives
• •Presented AI strategy to 150+ employees and advised on vendor evaluations

Python · llama.cpp · Weaviate · FastAPI · Streamlit · Llama 3