Alen Kabdislam

I’m a Computer Vision / OCR engineer and Team Lead with a strong Python background and production ML experience. I build end-to-end document analysis pipelines—from preprocessing and detection/segmentation to recognition, understanding, and API deployment. I obsess over model quality and inference performance with GPU-efficient utilization. At Jusan Bank, I lead DS/ML work end-to-end, architecting REST/FastAPI inference services with request validation, async processing, logging, and versioning. I drive batching, memory-aware inference, and throughput/latency trade-offs, while mentoring engineers on reproducibility, code quality, and production readiness. I’ve also built OCR & Document Understanding and YOLO-based detection/tracking pipelines, optimizing execution on limited VRAM through resizing, batching, and lightweight post-processing. From labeling guidelines and QA to augmentations and synthetic data, I create data workflows that translate into reliable structured outputs (JSON/Excel).