Chief Engineer, Robotics · MUSE Inc. · Tokyo, Japan
I am giving a soul to robots that work in real stores, and making them a little more reliable every week.
At MUSE Inc.
As Chief Engineer, Robotics at MUSE Inc., I lead the architecture of a new generation of store robots: a multi-purpose, modular AMR that takes on different jobs by switching units. It was built from the ground up and runs daily in live operations. Before that, I spent years bringing AI onto small, low-power devices.
As a student I took part in many robot competitions and won quite a few awards, yet every one of those robots stayed in the lab. I wanted to build robots that leave it and get used every day, and that is what I work on now.
That path taught me to think across the whole stack, from algorithms and hardware to the teams and systems that hold them together. I care about engineering that is simple, reliable and measurable, because a robot only matters when people can trust it. I work AI-first: coding agents are part of my daily team, and I set up the specs and checks that keep their output trustworthy.
I am happiest where robotics, embedded systems and AI meet, in a team that ships to real users and hears back from them quickly. If you are working on robot software, motion control, or on how engineers and AI agents can build things together, I would be glad to talk.
MUSE · 2026
Built end to end by our 5-person robot software team for a multi-purpose modular AMR: a layered Rust and Zenoh architecture where new units plug in through configuration, with one-step updates and automatic rollback.
MUSE · 2023—
MPC path tracking that keeps a round robot and its cart steady on straights and through corners, tuned on real-robot logs replayed in simulation.
MUSE · 2026
Guided by a rear camera reading marker boards, the robot backs up to a cart, docks, and tows it away. It releases the cart on its own too.
MUSE · 2026
An AI model predicts a safe, drivable path from fused LiDAR and camera data, running on an on-board NPU.
Renesas · 2022—2026
Test, profiling and benchmark tooling that shows how AI models run on resource-limited MCUs.
QuantumCore · 2020—2022
Lightweight signal processing and ML for radar, IR and audio on low-spec hardware.
MUSE Inc. · Tokyo
2026 — present Technical lead, next-generation robot system
2023 — 2025 Motion control and obstacle avoidance (approved side job while at Renesas)
Renesas Electronics · Tokyo
QuantumCore Inc. · Tokyo
I took on motion control and obstacle avoidance as an approved side job. The results led to being asked to lead the whole new system, and I moved over full-time.
Find the root cause, model it, replay real-robot data in simulation, and only then change the code. Give the team one number to move; for our store robot, the intervention rate. I hold Japan’s Statistics Certificate, Grade Pre-1.
From signal processing and embedded AI to robot control and system architecture, one thread runs through my work: making algorithms run on real hardware. Each step built on the last, and each shipped a product.
Coding agents are part of my daily team. I write the specs, tests and conventions that make their work reliable, and review what they ship.
Shared contracts, one-command environments, CI, and specs and decision records that keep the reasoning when people change, so everyone can build in parallel.
Every week I visit a store with the engineers. Time beside the robots shows problems no dashboard does.
Always happy to connect with people working on robotics, autonomy and embedded AI.
clinfeng.jlu@gmail.com