Computer Vision • Deep Learning • MLOps

Production-grade computer vision. Built to ship.

Detection • Tracking • Re-ID • 3D • Optimization (TensorRT / TFLite). I design, train, and deploy systems that run on edge and cloud.

10+Years in CV/AI
25+Deployed Models
Curiosity

Featured Projects

Real-time Multi-Object Tracking

Edge-optimized tracker (TensorRT) at 60 FPS on Jetson Orin. Robust to occlusion and low light.

Re-ID + Analytics for Retail

Person re-identification with privacy-aware embeddings and heatmaps for dwell/time-in-zone.

3D Pose for Robotics

Keypoint models + IK for grasp planning; sub-50 ms end-to-end with quantization and pruning.

Articles

Tracking under occlusion

Practical tricks for MOT stability: motion priors, re-ID hints, and temporal smoothing.

Toward shippable models

Quantization, TensorRT/TFLite paths, and what to profile before choosing hardware.

Monitoring CV in production

Input drift, alerting, and dataset refresh: an MLOps checklist for vision systems.

Services

Prototyping → MVP

Fast experiments, success metrics, and an MVP you can demo to stakeholders.

Optimization & Porting

Torch → TensorRT/TFLite, INT8 calibration, latency/throughput trade-offs.

Production Integration

APIs, streaming pipelines, monitoring, retraining loops, and on-device updates.

Contact