DSML.KZ Community Hub

TargetAI is focused on video analytics solutions. Responsibilities: • Designing and developing CV/Video Analytics models: object detection (person, vehicle, face, etc.), tracking, recognition (face, license plates, attributes) • Working with video pipelines: RTSP/HTTP - decoding, frame sampling, batching • Adapting and customizing models for real-world conditions: poor lighting, different camera angles, noise, compression, FPS drops • Fine-tuning and retraining models for specific customer scenarios • Optimizing inference for CPU • Reducing latency and resource consumption • Balancing accuracy vs performance • Integrating ML models into production: backend services, edge devices, on-prem installations • Monitoring model quality: precision/recall, data drift • Preparing and analyzing datasets: labeling, validation, data augmentation • Building training and testing pipelines • Analyzing model errors and quality degradation Requirements: • 4+ years experience in Computer Vision/Video Analytics • Strong understanding of CNN, YOLO, SSD • Experience with Face recognition pipelines • Knowledge of OCR/ALPR • Proficient in PyTorch or TensorFlow, OpenCV, ONNX • Understanding video formats and streams: RTSP, codecs (H.264/H.265), FPS, bitrate, latency • Experience optimizing CPU-only inference • Experience working with Linux • Deep understanding of OpenVINO • Knowledge of edge-device logic • C++ for performance-critical components • Experience with Kafka • Experience with gRPC/REST • MLOps skills: MLflow, model versioning, monitoring • Experience with large-scale CCTV projects