Ni Ni Tin Win
Building AI systems
I care about making AI reliable - data quality, evaluation, and the human layer that makes models trustworthy.
About
I co-founded a consumer app and scaled it from 0 to 2,000+ users, then joined an AI robotics startup as an operator and was promoted within months to lead the team, helping grow the data operation from 25 operators to 130 across 10 sites.
I think like a PM and build like an engineer. Recently: a full-stack voice AI agent shipped in a single day – FastAPI backend, 13 endpoints, 9 wired as agent tools across three live third-party APIs – and a matching engine whose tiered pipeline cut LLM calls 86%, validated with a custom eval harness.
Three threads, one job: turning messy problems into systems that work – a data pipeline, a team, or a product. I do my best work where people run with a whole problem, not just their slice.
If there's a problem to solve, I find a way.
Projects
Matching Engine
An AI-native, domain-agnostic compatibility-matching engine built on a hybrid scoring model - a deterministic core plus a bounded LLM nuance layer - designed for explainable scores and an eval-first workflow. Applied LLM security practices: prompt-injection mitigation, input/output validation, and secure handling of API keys. Interactive front-end in React & TypeScript.
Timeline
Made the full transition to building AI systems
Deepening AI engineering.
2026
AI Data Operations at a robotics company
Supported AI model training and robotics troubleshooting.
2024
Designed AI-adjacent product experiences
Designed a veteran admissions portal; front-end design for an AI insights platform.
2023
Began formal study in product and graphic design
Studied in San Francisco.
2022
Co-founded Chat-Chin, Myanmar's first car-servicing app
Grew the app to 2,000+ users.
2019