Mayur Hulke · Staff Autonomy Engineer at HIVE
One brain. Every machine. 🤖
Now
Learned autonomy for heavy machines
Staff Autonomy Engineer · HIVEHIVE retrofits the machines customers already own, from forklifts to wheel loaders, so they work under human supervision instead of a driver in the cab. I lead the learned-autonomy work: vision-language-action models, the data pipelines behind them, and one shared intelligence that learns from real work across machines and sites.
Operating in the field, not in demo mode · 40+ people across London, Norway and the USSo far, I've deployed
Factory perception to robot action
Robotics Lead · Almetra, seed to scale-upPre-trained vision-language models that turn factory video into timelines of the work, distilled to run on-station. Then built the robotics team that used that data to put Franka and UR arms on production lines.
60+ factory sites · 10,000 video-hours/day · Customer pilot with a robot in the loop · Grew to 40 peopleNon-intrusive roadside perception
Built & deployed · Q-FreeAn offline, camera-based edge system that recovered traffic counts, vehicle classes, speeds and flow across multiple lanes and changing weather.
99.7% detection accuracy · 93.5% mAP · 65+ FPS · 10 units deployed across the UK and US3D vehicle perception & inspection
Built & deployed · BCAA production vision stack for remote vehicle buying: scene transformation, reflection handling, 360° capture, multi-view 3D reconstruction and fine-grained cosmetic-damage detection.
Inspection time: ~4 hours → under 90 minutes · Millions of vehicle imagesAdaptive powered bionic leg
Founded & led · Chisel RoboticsA fully actuated lower-limb prosthesis using multimodal sensing and machine learning to personalise gait and adapt to changes in the user and walking surface.
Under 50 ms closed-loop feedback · 4 NHS trials · Performance legs for a Rio Paralympic gold medallistThe longer story
Working on learned autonomy for heavy machines? Message me on LinkedIn.