I build AI tools and data products for real-world decisions.
My work spans an internal AI assistant at Ourea Energy, a public-sector careers platform I co-founded, and research software. I’m pursuing an MS in Computational Analysis and Public Policy at the University of Chicago.
Ourea Brain is an internal AI assistant that brings company context into Slack, so colleagues at Ourea Energy can ask where a project stands, who owns it, and what was decided.
What I built
As an Agentic AI Intern, I built and deployed it: a Slack-connected application on Google Cloud Run that draws on Slack, Gmail, Google Docs, and meeting notes and uses Anthropic’s Claude API. It is designed to help colleagues retrieve project context, identify owners, and follow up on decisions.
Observed behaviour
An early live deployment answered a Slack question using meeting-note context.
Internal tool at Ourea Energy; no public link.
How it works (simplified architecture)
A colleague asks a question in Slack.
The app on Google Cloud Run receives it; credentials are held in Secret Manager. my work
It gathers relevant company context and calls the Claude API. my work
The response is posted back in Slack.
Tools: Slack integration, Anthropic Claude API, Google Cloud Run, Google Workspace sources, meeting-note integration, Secret Manager.
Making public-sector opportunities easier to find.
What it does
PublicPath is a platform that helps students and early-career candidates explore government careers.
What I built
I co-founded PublicPath and was one of two engineers. My work focused on API ingestion and data quality: cleaning, deduplicating, normalizing, and validating listings so they could be searched and filtered.
Project artifact
PublicPath homepage, captured 6 October 2026. This is the team’s product; it does not show my ingestion work, which is described above. Project website ↗ Job search was not loading when checked on 6 Oct 2026.How it works (simplified data flow)
Listings come from external job APIs.
Python ingestion is scheduled to run daily with GitHub Actions. my work
Data-quality steps clean, deduplicate, normalize, and validate listings. my work
Listings are stored in Postgres on Supabase.
The site’s search and filters read from that store.
Turning clinical notes into patient-friendly explanations.
What it does
CLARITY is a healthcare-AI research project; its pipeline turns de-identified clinical notes into patient-friendly explanatory videos.
What I built
I built the Python pipeline. It takes de-identified clinical-note JSON and automates the explanatory content, narration with Microsoft Neural TTS, slide generation, and MP4 production.
Output
MP4 explanatory videos.
Clinical material is not public, so no example is shown here.
How it works (pipeline sequence)
Input: a de-identified clinical note in JSON.
Explanatory content is produced from the note. my work
Narration (Microsoft Neural TTS) and slides are generated. my work
