Daniel Yamin

I build LLM-powered applications and agentic workflows for clinical genomics at Franklin by QIAGEN, where I’m the go-to technical resource for AI development across the organization.

Before that I spent a decade shipping production systems — DeFi risk platforms at Chaos Labs, IoT security at Microsoft Azure, real-time network anomaly detection at Allot — and completed an MSc in computational neuroscience at Tel Aviv University, where I co-developed a way to measure human memory from eye movements alone, published in Nature Communications Psychology.

I studied how brains remember. Now I build systems that give AI agents memory, tools, and judgment.

What I do

From human memory to machine memory

My research asked a strange question: how do you detect a memory the person can’t tell you about? Our answer was to read it from the eyes — anticipatory gaze reveals a memory trace seconds before the remembered event appears on screen, even when people say they don’t remember.

Building agentic systems raises the engineering version of the same question: what should an intelligent system remember, and how do you prove it remembered? Memory, evaluation, and observability for AI agents aren’t buzzwords to me — they’re the problem I studied in the lab, now in production form.

Experience at a glance

Education: MSc Computational Neuroscience, Tel Aviv University (GPA 98) · BSc Computer Science & Cognitive Science, Open University of Israel (GPA 92, honors track scholarship)

Open source

The MEGA research implementation is open source (MIT), together with its experiment presentation suite — the full pipeline behind the paper, from stimulus presentation to gaze-based memory classification.

Let’s talk

I’m always happy to hear about hard problems at the intersection of AI, data, and product — collaborations, speaking invitations, or a problem you think I’d enjoy.