Breaking
MIN LEE named 2025 Visionary CEO in Emotional AI EchoAI reports 96% emotional comprehension rate 130,000+ training hours since 2016 IPMD recognized as a Google Technology Build Partner Founded at UC Berkeley · built in San Mateo
Founder · CEO · IPMD, Inc.

Min Lee

He spent more than 30,000 hours teaching a machine the difference between a polite smile and a real one. The company is IPMD. The engine is EchoAI. The bet is that your face is the next interface.

Emotional AI Micro-expressions San Mateo, CA Since 2016
Min Lee, founder and CEO of IPMD, Inc.
Min Lee. The face that studies faces.
// The Pitch

A second opinion on a feeling you haven't named yet

Most software waits for you to type, tap, or click. Min Lee built something that watches instead. EchoAI - the platform his company IPMD calls Project M - looks at a face, listens to a voice, reads the words, and within seconds hands back a report on what the person is actually feeling. Not what they said. What the micro-expression gave away.

That is a strange thing to spend a decade on. It is also the thing Min Lee, founder, CEO and president of IPMD, Inc., has spent a decade on - from a project that started on the UC Berkeley campus in 2016 to a San Mateo company recognized in 2025 as a Google Technology Build Partner. The team is small, around 33 people. The ambition is not.

Technology should enhance human connection, not replace it. - Min Lee, founder & CEO, IPMD
96%
Emotional comprehension rate
130k+
Hours of training data
~500
Collaborators on the build
// What He's Building

Project M skips the muscles and hunts the flicker

The approach

Not your standard computer vision

Most emotion-recognition systems track facial muscle movement - the obvious tightening of a brow, the pull of a lip. Project M deliberately walks away from that. It chases the hidden and micro facial expressions, the ones that surface for a fraction of a second before the mask goes back up. That design choice is the whole thesis: the real signal is the one people try to hide.

The output

A digital mirror

EchoAI reads facial expressions, vocal tone, language, and micro-expressions together, then returns a personalized emotional analysis report - a reflection of states a user may not even realize they are feeling. A planned premium tier pairs the engine with human emotional coaches. The platform lives at echoai.ipmdinc.com.

Receipts: EchoAI reports a 96% ROC/AUC score and consistently out-benchmarks emotion models from major technology providers. It was trained on 130,000+ hours of multimodal emotional interaction data collected between January 2016 and December 2023.
// By The Numbers

Ten years, measured

EchoAI emotional comprehension96%
Scientific ROC / AUC score96%
Min Lee's share of total training hours (~30k of 130k)~23%

Figures as reported by IPMD and Business Worldwide Magazine, 2025.

// The Arc

From a campus project to a build partner

The detail that explains Min Lee is the 30,000 hours. Not the title, not the award - the hours. Of the roughly 130,000 hours that went into mapping and labeling human emotion for EchoAI, he personally logged more than 30,000 of them. Founders usually delegate the grind. He sat inside it.

IPMD took shape at UC Berkeley in 2016, and the build leaned on a rotating team of nearly 500 students, drawn largely from the same campus. For seven years - 2016 through 2023 - that group did the unglamorous work of teaching a system what an emotion looks like when it is trying not to be seen. The result was Project M, the emotion engine, and later EchoAI, the product wrapped around it.

When the model was unveiled, the claim was blunt: emotion recognition accurate enough to rival the industry's biggest names, built by a company most people had never heard of. The approach got attention precisely because it refused the standard playbook. Instead of mapping facial muscles, it went after the flickers.

By 2025 the recognition caught up. Business Worldwide Magazine named Min Lee a Visionary CEO in Emotional AI & Human-Centered Healthcare Innovation on September 30, 2025. IPMD landed Google Technology Build Partner status. The company that started in a campus lab now builds agentic AI systems and partners on next-generation platforms - all from a 33-person operation in San Mateo.

// Timeline

The long build

2016

Founds IPMD, Inc. at UC Berkeley and starts work on an emotional-AI platform.

2016 - 2023

Leads a 130,000+ hour emotion-mapping effort with roughly 500 collaborators, contributing 30,000+ hours himself.

2023

Project M unveiled - emotion recognition reported at ~96% accuracy, rivaling major providers. Seed funding round.

2025

EchoAI launches. Named Visionary CEO in Emotional AI by Business Worldwide Magazine (Sept 30). Recognized as a Google Technology Build Partner.

// The Person

Quirks, receipts & one good line

Lean by choice

About 33 employees, yet a Google Technology Build Partner. Min Lee runs a small shop that punches at the weight of much larger labs.

The Berkeley engine

Much of EchoAI's emotional model was built by a rotating crew of around 500 students - a campus turned into a training-data factory.

An Otherdots advocate

Selected as an Advocate within the Otherdots Foundation network, opening Project M to startups building on top of it.

Otherdots Foundation and IPMD are partnering together, which allows the OD startups to use the power of M in different applications. - Min Lee

Watch: Meet Min Lee and Project M, partnering with Otherdots →

// Filed Under
emotional aiemotion recognitionechoaiproject m micro-expressionshuman-centered aiuc berkeleyaffective computing agentic aifounderdeep learningsan mateo