MoodSense AI
SHIPPEDIndependent
Fine-tuned DistilBERT for emotion classification
Fine-tuned DistilBERT for emotion classification, reaching 92.95% accuracy.
→ 92.95% accuracy
DistilBERTTransformersPython
Work
Three systems with full write-ups — architecture, the decisions I got to make, and what I rejected. Everything below those is shipped work already running for real teams.
Independent
Production AI agent platform — plans, acts, and verifies its own work
Getting a computer to actually finish a multi-step task across your email, your notes, and the web still means doing every step by hand.
Skillwyn Labs
Technical due-diligence platform — the audit a VC associate does by hand
Before anyone invests in or buys a company, someone has to check whether the product is actually as good as the pitch — and today that's a person clicking through it by hand for days.
Independent
Evidence-first verification — every conclusion traces to evidence, a human decides
Hiring tools hand you a score for a candidate without showing you where it came from, so you're trusting a number you can't check.
Production work at Stirring Minds and Skillwyn Labs, plus earlier model work.
Independent
Fine-tuned DistilBERT for emotion classification
Fine-tuned DistilBERT for emotion classification, reaching 92.95% accuracy.
→ 92.95% accuracy
Stirring Minds
Automated resume screening against real job descriptions
Parses incoming resumes and scores each candidate against the job description using retrieval-augmented ranking rather than keyword matching. Cut manual screening time by roughly 70% for the recruiting team using it.
→ ~70% reduction in manual screening time
Stirring Minds
Autonomous voice agent for Tier-1 sales and renewal calls
Built on Vapi — handles live phone conversations end-to-end, including objection handling and negotiation, on Tier-1 sales and renewal calls without a human on the line. Deployed to real customers.
Skillwyn Labs
Turned a fully manual outreach workflow into a zero-touch pipeline
A lead lands in a Google Sheet and the pipeline takes it from there — OpenAI drafts outreach personalized to that specific contact, and n8n sends it through Gmail, with no manual drafting or sending in between. Built and shipped for a client running outreach at volume, where the manual version didn't scale.
Stirring Minds
Answers support queries from live policy and location data
A retrieval-backed support assistant that answers customer questions by grounding responses in current policy documents and location-specific data instead of a static FAQ.
Autonomous AI execution system — early exploration, scope still being shaped.