About
I build agents for decisions that involve more than one person, and I make them explain themselves.
Most of what I build is about a decision that more than one person has a stake in: where a group should meet (MeetSpot), how two people settle a deal (Negotiation Arena), who is responsible when an agent gets it wrong (SafeReceipt). Each one has to show why it chose what it chose.
Experience
MindLight Tech, AI Engineer Intern
Apr 2026 to now, remoteDeepWisdom, Agent Algorithm Intern
Oct 2025 to Apr 2026, remote- Built an SEO agent that pulls from 7+ data sources (Google Trends, GDELT and others), runs 150+ technical checks, and applies fixes itself. Core page SEO scores went from 60 to 80.
- Built a research-report module on RAG that combines sentiment analysis with BLS labor data and writes 2,000 to 3,000 word market briefs.
- Moved slow API calls into a local-script-plus-webhook setup, wired to Supabase and Zapier, so the content pipeline stopped hitting rate limits.
Research
QualityAgent, first author, IEEE IRAI 2026 (Melbourne). Three agents turn a photo of a manufacturing defect into CAPA and PFMEA documents that follow ISO 9001, and a human supervises what they produce. The paper received the IEEE IES SYPA Travel Award.
Awards
- IEEE IES SYPA Travel Award, 2026
- 2nd Prize, Soul App Agent Hackathon Final (MeetSpot)
- Best Topic Award, OpenManus Hackathon (VibeDoc)
- Excellence Award, 12th Nvidia Sky Hackathon (WeFinance)
- Star of Tomorrow Special Award, L’Oréal Hackathon
The full list of builds is in the hackathon log.
Tools
Python (FastAPI, LangGraph), TypeScript, React, Postgres, Docker. I contribute to the Datawhale and WaytoAGI communities.
Contact
Email is fastest: johnrobertdestiny@gmail.com. I’m open to agent engineering roles, contract work, and open-source collaboration. Also on GitHub, X and LinkedIn.