Skip to content

Resume

Sidharth Kudupudi

Engineer · Minneapolis, MN

Selected projects

  • Infrared small-target detection & tracking

    1.7 ms per 1280×1024 frame, enhance → detect → track

    I built a detector and tracker for 3–7 pixel aerial targets in infrared video, then ported it from Python to C++/CUDA/TensorRT: 1.7 ms per frame, 9.3× faster, with tracks identical to the prototype.

    PyTorch · YOLOv8 · CUDA · TensorRT · C++17 · Eigen

  • I built an interpretable pipeline on real cobot telemetry: an unsupervised detector flags faulty cycles at AUROC 0.944 with a 1.7% false-alarm rate, and a Random Forest names the fault.

    pandas · scikit-learn · PCA · Random Forest · NumPy

  • I built LiDAR-to-camera projection for a 120° f-θ lens from raw NVIDIA AV files and gave every segmented object a metric range, over two 605-frame drives.

    Python · NumPy · DracoPy · rosbags · ROS 2 · YOLO11-seg

  • Industrial defect segmentation with a C++ TensorRT app

    11 / 15 held-out defects located by the C++ app

    I trained YOLO11-seg on MVTec AD to 0.82 mask mAP50, wrote the C++ TensorRT app that serves it, and found that rebalancing VisA's data lifted mask mAP50 from 0.25 to 0.41 where a bigger model did nothing.

    Ultralytics YOLO11-seg · TensorRT · C++17 · CUDA · OpenCV · ONNX

  • Sub-pixel Fourier registration for wafer metrology

    0.015 px mean error on a DRAM array

    I implemented phase correlation with matrix-multiply DFT upsampling in NumPy: 1/100-pixel shifts that match scikit-image exactly on 50 of 50 tests, benchmarked across four wafer patterns and 15–40 dB of noise.

    NumPy · SciPy FFT · Phase correlation · scikit-image (cross-check)

  • I validated a surgical-tool detector at 0.963 mAP50 over 5 folds, deployed it as a 0.87 ms TensorRT engine behind a Qt panel and a C++ serial relay, and measured an honest cross-dataset drop to 0.17.

    YOLO11n · TensorRT · ONNX · PySide6 · C++17 · POSIX termios

  • I wrote a SORT tracker from scratch, ran it against BoT-SORT with camera-motion compensation on all 17 VisDrone test sequences, and caught an ID-leak bug through the output analysis.

    RT-DETR-L · Kalman filter · Hungarian assignment · BoT-SORT · motmetrics · Ultralytics

  • I fine-tuned Qwen2-VL-2B with QLoRA to 100% valid JSON, then showed on held-out clips that its intent decision stayed at the majority baseline, traced why, and measured what a VLM does to a real-time tracker.

    Qwen2-VL-2B · QLoRA (4-bit NF4) · PEFT · YOLO11s · PyTorch · Python threading

  • I scored a Grounding DINO + SAM zero-shot pipeline against 744 annotated plumes, raised localisation from 36% to 51% through prompt design, and put the MobileSAM encoder on TensorRT at 4.3 ms.

    Grounding DINO · Segment Anything · MobileSAM · TensorRT · ONNX · PyTorch

  • Multi-camera BEV perception: Lift-Splat with DINOv2

    6 cameras lifted into one 200×200 bird's-eye-view grid

    I built a Lift-Splat bird's-eye-view network with a DINOv2 foundation-model backbone, module by module, and verified its six-camera geometry end to end on real nuScenes calibration.

    PyTorch · DINOv2 ViT-S/14 · Lift-Splat-Shoot · nuScenes devkit