Curriculum Vitae

London, United Kingdom

• Contributed to the company's growth from early prototype to securing over $2.5M in venture funding.
• Led the design and development of a real-time computer vision system for surveillance and unmanned system automation. Architected an event-driven, lock-free MPMC double-buffer pipeline to optimize latency and minimize memory locality issues — achieving 6 simultaneous 1080p camera streams on NVIDIA Jetson and 40 streams across 4 NVIDIA Ada RTX 6000.
• Built and optimized pipelines for multiple deep learning models, including object detection (DETR/YOLO), object tracking (SiameseRPN++, DeepSORT), pose estimation (AlphaPose), and face redaction. Trained on GCP with PyTorch/TensorFlow; deployed with TensorRT for high-speed inference.
• Led performance optimization across the stack (profiling, memory locality, batching, async execution); accelerated critical paths by ~3× (measured in Nsight Systems / Remotery) by moving CPU stages to CUDA kernels without compromising model accuracy.
• Developed a low-latency WebRTC streaming server using GStreamer and Node.js to deliver real-time HD video to browsers and RTSP endpoints.

• Designed a two-stage infrared anti-spoofing system, achieving < 3% false acceptance rate (FAR).
• Improved the face recognition model using knowledge distillation and network optimization — delivering a 4× speed improvement with equivalent accuracy.