I build agentic AI systems for a living. Not the kind you see in demo videos — the kind that run 24/7 in enterprise supply chains, where a bad decision means a container ship goes to the wrong port.
For the past 10+ years, I’ve worked across the full spectrum of AI/ML: from publishing research on optimization and machine learning, to leading teams that architect and deploy multi-agent platforms at enterprise scale. I hold a PhD in Computer Science, and I currently serve as Data Science Manager at a tech start-up company.
Why this site exists
I started harness-engineering.ai because the gap between “agentic AI as discussed on Twitter” and “agentic AI as deployed in production” is enormous. Most content about AI agents is written by people who’ve never had to answer the question: “Why did your agent autonomously reorder $2M of inventory at 3 AM?”
This site covers the engineering reality of building agentic systems:
- Architecture patterns that survive contact with production data
- Orchestration frameworks for multi-agent systems that need to be debuggable, not just impressive
- Evaluation and observability for systems where you can’t just eyeball the output
- Failure modes I’ve personally encountered (and the scars to prove it)
Everything here comes from building real systems. If I haven’t deployed it or seen it deployed, I won’t write about it.
Background
- PhD in Computer Science
- 10+ years building AI/ML systems, from research to production
- Currently leading data science at a tech start-up serving the S&P 500 enterprise client.
- Published researcher in optimization and machine learning
- Architected agentic AI platforms processing autonomous decisions at scale
Get in touch
For consulting inquiries about agent infrastructure, framework selection, or production readiness reviews, reach out here.
For weekly deep dives on agent harness engineering, subscribe to the newsletter.