AI Parking (aiparking.kz)
up to 1,000,000 KZT / month (before tax)
Hybrid, Almaty · Full-time · 2-month probation period
AI Parking is a parking automation platform: license plate recognition, cameras, boom barriers. We are launching a major project: cameras monitoring people.
Responsibilities:
- Develop and improve license plate recognition (ANPR/LPR) models: night, dirt, glare, angle, Cyrillic/Latin characters
- Process RTSP/IP streams in real-time
- Reduce recognition errors (currently up to ~10% on some objects) — FP/FN, character confusion, false entries/exits
- Collect and label datasets, clean, augment
- Train and fine-tune models (PyTorch), optimize inference for edge/on-prem
- Develop models for human surveillance: detection, tracking, action recognition
- Work with quality on live objects, not just on benchmarks
Requirements:
- Strong middle / senior
- Strong Python
- PyTorch (TensorFlow is a plus)
- CV: detection, tracking, OCR, classification
- OpenCV, working with video and RTSP
- Understanding of training: loss, overfitting, augmentation, metrics
- Production inference experience: latency, throughput, TensorRT/ONNX/OpenVINO is a plus
- ANPR/LPR or OCR experience is a strong plus
- Human detection / tracking / action recognition experience is a plus
Will be a plus:
- Nomeroff-net, YOLO, PaddleOCR
- Edge: NVIDIA Jetson / GPU servers
Conditions:
- Report to the Development Team Lead
- Company-provided equipment
