Expert in Deep Learning & Computer Vision, including CNNs, Vision Transformers, YOLOX/YOLOv7/YOLOv8, PointNet/PointNet++, VoteNet.

Strong in LLMs & Generative AI, including GPT/Llama fine-tuning, RAG pipelines, Qwen2-VL, Stable Diffusion/diffusion models.

Advanced backend & MLOps: Docker, FastAPI, Streamlit, AWS (S3, SageMaker), Airflow, Linux, optimisation with TensorRT/DeepStream.

Skilled in Python, C++, C#, Java, Golang, building production-grade ML systems across detection, forecasting, NLP and optimisation.

Delivered large-scale detection systems achieving 96% accuracy with <1% false positives, processing 50k+ images across multi-camera streams.

Achieved 90–95%+ performance across 3D CV tasks (classification, segmentation, detection) using PointNet/PointNet++ and VoteNet.

Drove commercial impact through multimodal/LLM solutions (RAG, Qwen2-VL fine-tuning, GPT/Llama training), enabling automation and faster decision-making.

Built and deployed real-time, edge-optimised pipelines (TensorRT, DeepStream) enabling automated vehicle/pedestrian analytics and operational efficiency gains.