About
Sidharth Kudupudi
Engineer · Minneapolis, MN
I'm an engineer with an M.S. in Robotics from the University of Minnesota, originally trained in mechatronics. I like the whole arc of a problem: making sense of messy data, building something that works, then making it fast and dependable enough that someone else can rely on it.
I care about evidence more than demos. I test on data a system has never seen, put the unflattering number next to the flattering one, and write down what didn't work.
Skills, grouped by where I use them
- Perception
- Detection & segmentation · Multi-object tracking · LiDAR–camera fusion · BEV perception · Vision-language models
- Deployment
- C++17 · CUDA · TensorRT · ONNX · pybind11 · CMake · Latency profiling
- Estimation
- Kalman filtering · Hungarian assignment · Camera calibration · Phase correlation
- Robotics
- ROS 2 · Isaac Sim · Gazebo · RViz · URDF
- ML
- PyTorch · Ultralytics · Hugging Face · scikit-learn · LoRA / QLoRA
- Data
- nuScenes · NVIDIA PhysicalAI-AV · VisDrone · MVTec AD · JAAD · voraus-AD
How I work
01
Measure on data the model never saw
Held-out sequences, cross-dataset tests and variant-held-out folds, reported next to the flattering number.
02
Report the trivial baseline
Accuracy means little without the majority-class score beside it; false alarms belong next to recall.
03
A port must match its prototype
Stage-by-stage parity tests before quoting a speed-up.
04
Say where it fails
Every case study has a failures section with real examples, not just limitations in the abstract.
Contact
Got an interesting problem? Let's talk.
Open to full-time engineering roles. If the problem is hard and the data is messy, I'm interested.