About this role
Summary: Join the R&D team to develop and deploy machine learning models for on-edge inference in video telematics devices. Responsibilities include model selection, optimization for resource-constrained hardware, and integration with embedded systems.
We are looking for AI Engineer to join our R&D team in Warsaw, Poland. You will join one of the teams developing our next-generation video telematics devices and be responsible for analyzing, selecting, testing, and deploying machine learning models used for on-device (edge) inference, PC-based validation, and — going forward — cloud-based processing of data collected from our devices. We'd love to walk you through all the details during the recruitment process!
Teltonika’s Telematics business unit develops and produces easy-to-use GPS tracking devices for effective fleet management across logistics, agriculture, car sharing, construction, and more. We use the latest technologies to set new standards for the telematics industry.
If you want to shape how AI models run at the edge on real embedded hardware, not just in a notebook, this is the team for you!
In this role, you will:
- Analyze, select, and evaluate machine learning models for specific use cases and problem-solving needs (e.g. ADAS/DMS, video/image analysis, sensor and CAN data processing).
- Test, benchmark, and configure AI models for deployment on embedded edge devices, as well as on PC (Windows, Linux) platforms.
- Analyze system capabilities (CPU, GPU, NPU, memory, power budget) and contribute to selecting hardware of appropriate specification for AI workloads.
- Optimize models for resource-constrained edge hardware (quantization, pruning, conversion to inference runtimes such as TensorFlow Lite, ONNX Runtime, or vendor-specific SDKs).
- Design the data flow between edge-level pre-processing and any future cloud-based models that further process data collected from edge AI devices.
- Collaborate with embedded software and hardware teams to integrate models into the media/data pipeline of our video telematics devices.
- Track current and emerging AI/ML trends and evaluate their applicability to company products.
- Document model performance, testing methodology, and deployment decisions.
- Participate in Agile/Scrum team processes and decision making.
What we expect of you:
- 2+ years of experience in AI/ML model development, evaluation, or deployment.
- Solid Python skills and practical experience with ML frameworks (PyTorch, TensorFlow, or similar).
- Experience in testing and benchmarking models for accuracy, latency, and resource usage.
- Understanding of the constraints of running inference on embedded/edge hardware (CPU/GPU/NPU limitations, memory, power).
- Comfortable working across both Linux and Windows environments.
- Strong analytical skills and the ability to independently investigate and solve technical problems.
- Professional working proficiency in English.
Ideally, you’ll also have:
- Experience with edge inference runtimes and toolchains (TensorFlow Lite, ONNX Runtime, TensorRT, or vendor NPU SDKs such as NXP, Qualcomm, Mediatek, Rockchip RKNN, or similar).
- Experience with computer vision models relevant to ADAS/DMS (object detection, driver monitoring, event classification).
- Experience with C/C++ for integrating models into embedded applications.
- Familiarity with embedded Linux and camera/video pipelines (V4L2, GStreamer).
- Experience with cloud ML infrastructure (model serving, MLOps pipelines, containerized inference).
- Bachelor's degree in Computer Science, Data Science, Electronics Engineering, or a related field.
- Any other relevant experience that you think makes you a great fit!
What we offer:
- Private medical care (Medicover Premium) and a MultiBenefit cafeteria
- Unique opportunities to grow professionally being part of a challenging and high-growth environment
- Interesting and ambitious projects of industrial GPS tracking devices
- Additional day off for your birthday
- Team building activities