TECHNICAL PM — DATA & AI · TOKYO & US · EN / JA / KO

Hi, I'm
Jaeyoon.

Technical product leader with 7+ years across product management and data science, building analytics-heavy products.

Portrait of Jaeyoon
ABOUT

I'm a product manager and data scientist based in Tokyo. I was born in Korea, raised in Japan, and went to university and started my career in the US. Moving across cultures shaped how I think about products, communication, and people.

I started my career as a data scientist before gradually moving into product management, where I could get closer to customers and real-world problems. That data science foundation still shapes how I work as a PM: it helps me ask better questions, structure ambiguity, and build product intuition from evidence rather than assumptions.

Previously, I spent six years at Flatiron Health across data science and product management — building real-world oncology data products from unstructured EHR data, running experiments to optimize and forecast data operations, and launching the company's first ex-US data products in Japan, the UK, and Germany. In mid-2025, I left to build my own startup, Conatus AI. After many pivots we landed in the AEO/GEO space (AI search optimization) and reached paying customers at launch. Ultimately I made the call to wind it down in early 2026 — the traction was real, but I couldn't see myself committing the next five years to marketing tech.

Right now I'm taking my first real break in seven years, and figuring out what's next. AI × health is the obvious fit given my background, but the environment matters more to me than the domain: I'm open to anything, as long as it's an AI-native team where I can move fast and build things with intensity. I'm relocating to the US soon — see you on the other side!

FEATURED PROJECT

Juno: an AI wedding planner you can actually trust

One of my big projects during the break is planning my own wedding — so naturally, I built Juno, an AI agent for wedding planning, solo in about a month. Give it a shot if you have a wedding coming up — or read about how I built it.

Juno Today page
WRITING
Juno: an AI wedding planner The product case study — the design decisions and trust boundaries behind an AI agent that couples hand real work to. Evals instead of vibes Quality infrastructure for an agent product: evaluate the production code path, score deterministically first, diff against committed baselines, and grow test cases from production failures. Steering AI coding agents effectively The spec-driven workflow that built Juno — where it broke down, why CI was the game changer, and what agent teams taught me about product work.
CERAMICS
I occasionally make things on a wheel.
Ramen bowl Vase, French White & Orange Mini plates, Amethyst Thin neck vase Plate, Transparent & Blue Blue in green vase