CV/ML Engineer focused on optimized inference, model deployment, and high-performance computer vision systems.
Building scalable ML services with C++, Python, Docker, and ONNX Runtime.
| Project | Stack | Description |
|---|---|---|
| 🔥 yolo-pose-cpp | C++, ONNX Runtime, OpenCV, CUDA, Docker | High-performance pose estimation pipeline. Optimized from 10 to 94 FPS using GPU inference, batching, and CUDA providers. |
| Image Segmentation API | FastAPI, Docker, YOLOv8, ONNX Runtime | Containerized REST API for instance segmentation using FastAPI & Docker. |
| Triton YOLO Segmentation | Triton Inference Server, Docker, ONNX | Implementation of model serving using NVIDIA Triton, featuring model repository setup and client-side inference. |
| Multi-Model Detection Bot | aiogram 3.0, YOLOv8, Detectron2, MiDaS, ONNX | Telegram bot integrating object detection, segmentation, and depth estimation models. |
| E-commerce Analytics Platform | PostgreSQL, Docker, Airflow, FastAPI, PySpark | End-to-end data platform with ETL pipelines, ML forecasting, and interactive dashboards. |
Selected high-performing notebooks demonstrating deep learning expertise:
- PyTorch Lightning ⚡ | Xception | Test Acc: 99% — Image classification with custom preprocessing and Lightning training loop
- PyTorch Lightning | Baseline (no background) — Advanced data preprocessing techniques for CV tasks