Every year, road crashes kill 1.2 million people.
We're building the AI to stop it.
We’re building the foundation model for mobility — the AI that learns the physics of how people drive from sensor data across 65 million drivers worldwide. To date, we’ve prevented 126,000 crashes and 68,000 injuries. We’re just getting started.
will be built here.
The next decade of mobility AI will be built here.
The people who join CMT now — engineers, researchers, product leaders, designers, the teams who build the platform — will define how mobility AI works for the next decade.
Ours prevents them.
Most AI predicts outcomes. Ours prevents them.
Our foundation model for mobility learns the physics of how people drive — force, motion, trajectory — from real-world sensor data. It explains not just what happened on the road, but why. It can help everyone on the road better understand risk. More importantly, it can help reduce it.
There’s a lot more to build.
drivers worldwide
programs in production
crashes prevented
injuries prevented
It takes the right people to change the world.
Co-Founder & CEO
From safer drivers, to faster emergency response, to smarter infrastructure, we’re improving road safety from every angle.
foundation model
A real foundation model takes three things: hard problems, real-world data at scale, and a team that can build it from scratch. We have all three.
Atlas is a family of time-series transformers trained with self-supervised objectives on multi-modal sensor data — phone, vehicle, and IoT — to learn shared embeddings of driving dynamics.
The training data is proprietary, structured, behavioral — drawn from phones, vehicles, and CMT hardware running on real roads.
CMT powers the safe driving programs for the world's largest insurers, automakers, and mobility providers.
Risk measurement, crash detection, and life-saving safety programs — CMT’s technology aligns drivers and mobility providers around the same goal: safer roads, fewer crashes, faster help when it matters most.
The AI team
SVP of Artificial Intelligence
- Leads DriveWell Atlas and AI at CMT
- Ph.D. in Pure Mathematics from MIT
- Years applying mathematics and computation to high-stakes problems
- Holds more than a dozen patents
Co-Founder & CTO
- MIT Fujitsu Professor of Computer Science and AI
- 2023 Marconi Prize winner
- Recognized by The National Academy of Engineering & American Academy of Arts and Sciences
- Co-led the MIT CarTel project that became CMT
Co-Founder & Chief Scientist
- MIT College of Computing Distinguished Professor & Head of Computer Science
- Leads MIT's Data Systems and AI Lab
- ACM Fellow
- Co-founded Vertica (acquired by HP)
- Co-led the MIT CarTel project that became CMT
The playbook doesn't exist yet.
The problems are hard. The bar is high. Your model runs in production, your decisions show up on real roads, and the impact is real. That’s the work. The environment is built so you can do it well: small teams, direct access to leadership, engineers who’ll carve out an hour to debug with you because they care about getting it right. Work with a serious budget for compute and get runway to run real experiments.
How we work
Read the papers
Ask the next question
Run the experiment
Share what you learn
The best idea wins, regardless of who brought it
Engineer
Architect models that reach production-scale.
AI Researcher
Academic instinct, applied environment.
AI Leader
Own the foundation model roadmap.
Builder
Own a product end-to-end.
Storyteller
Shape how CMT shows up.
Partner
Help our customers and CMT grow.
Problem-solver
Deliver what customers need.
Operator
Run the engine.
People person
Build CMT from the inside.
We ask a lot.
We help you deliver.
What we build saves lives.
Competitive base, performance bonus, and RSUs in a profitable, growing company.
Unlimited PTO that people actually use.
Memorial Day through Labor Day.
80% medical, dental, and vision coverage for you and your family.
401(k) with company match.
16 weeks paid leave for primary caregivers, 4 weeks for secondary.
$3,600/year for education and learning.
Free lunch three days a week, commuter benefits.
Born out of MIT in 2004.
The idea for CMT started in 2004 with two MIT researchers exploring mobile sensing, processing, and communication — the three pillars that make modern telematics possible. Today, CMT is headquartered just down the street from where it all began — still on the MIT campus, still solving the hardest problems in mobility.
