Users create and collaborate with 3 AI agents in one place, and it cut task completion time by 40%.
Our RAG pipeline linked clinical entity data through BERT models across 3 hospital systems.
As models got more capable, I built agents and dev tools outside work, then shipped AI products professionally.
I treat every hidden condition as a design problem, so blocked, review, and recovery states get their own screens.
I fixed Windows cache-scan false corruption warnings in the Hugging Face Hub and ADLS path parsing in Airflow.
My clinical NLP pipelines pulled entities out of unstructured medical records for healthcare clients.
That research shaped how I reason about technical uncertainty and data-heavy product problems.

I build AI-powered products and data platforms as a Product Engineer.
Users create and collaborate with 3 AI agents in one place, and it cut task completion time by 40%.
Our RAG pipeline linked clinical entity data through BERT models across 3 hospital systems.
As models got more capable, I built agents and dev tools outside work, then shipped AI products professionally.
I treat every hidden condition as a design problem, so blocked, review, and recovery states get their own screens.
I fixed Windows cache-scan false corruption warnings in the Hugging Face Hub and ADLS path parsing in Airflow.
My clinical NLP pipelines pulled entities out of unstructured medical records for healthcare clients.
That research shaped how I reason about technical uncertainty and data-heavy product problems.

I build AI-powered products and data platforms as a Product Engineer.
Users create and collaborate with 3 AI agents in one place, and it cut task completion time by 40%.
Our RAG pipeline linked clinical entity data through BERT models across 3 hospital systems.
As models got more capable, I built agents and dev tools outside work, then shipped AI products professionally.
I treat every hidden condition as a design problem, so blocked, review, and recovery states get their own screens.
I fixed Windows cache-scan false corruption warnings in the Hugging Face Hub and ADLS path parsing in Airflow.
My clinical NLP pipelines pulled entities out of unstructured medical records for healthcare clients.
That research shaped how I reason about technical uncertainty and data-heavy product problems.
By using this service, you agree to the Terms of Service and Privacy Policy.