Andrey Kurenkov

Transitioned assembly line tracking from single- to multi-camera YOLO setup, cutting false positives by 19%. Built augmentation pipeline boosting mAP 0.72 to 0.77 with 20% synthetic data. Integrated Mistral 7B NLP interface for defect logging (70% adoption). Optimized models via TensorRT INT8 quantization, achieving 15ms latency (<5% mAP loss) on Jetson AGX Orin.

• Presented findings at SPIE Photonics West 2024, CYTO 2025, IEEE SMC 2022. • Adapted ML models for optical aberration correction and depth-of-field enhancement in imaging cytometry. • Developed CNNs and LLMs analyzing biomedical (blood cells, CT, X-Ray) datasets and patient data. • Optimized Brain-Computer Interfaces with subject-independent motor imagery classifiers, reducing calibration. • Conducted statistical analysis and literature reviews for ML in cell biology.

Integrated and load-tested eight new REST endpoints for a B2B (financial advisor firm) analytics feed, handling 25k daily requests during beta-stage at 99.8 % uptime. Added Grafana dashboard that tracked P95 latency regressions (>900 ms) in real time.