Embedded Vision Engineer — NPU AI for Autonomy (Delft)

The Vydar Group; Delft; On-site; EUR 70,000 - 110,000; Full time

Job summary

The Vydar Group in Delft is seeking a Machine Vision Engineer to design, implement, and optimize vision algorithms for constrained embedded hardware, focusing on NPU accelerators supporting autonomous defense platforms.

You will train detection, tracking and segmentation models, quantize and compile networks, profile latency and power, and collaborate with software, AI, electrical and mechanical engineering teams to deliver real-time, safe, and reliable perception stacks.

Qualifications

• NATO citizenship is required.
• MSc in Electrical Engineering, Computer Engineering, or equivalent.
• Hands-on experience with embedded hardware (MCUs, SoCs, DSPs, or FPGAs).
• Proven NN optimization for target hardware: quantization, pruning, fusion.
• Strong programming in Rust, C/C++, and Python for tooling.
• On-site work in Delft and clear technical communication.
• Strong problem solving and emphasis on safe, reliable design.

Responsibilities

• Design, implement, and optimize machine vision algorithms for embedded hardware.
• Train detection, tracking, and segmentation models for NPU accelerators.
• Quantize and compile networks; profile latency, memory, and power budgets.
• Collaborate with software, AI, electrical, and mechanical engineering teams.
• Contribute to data collection and annotation pipelines, on-vehicle integration, and field testing.
• Document and continuously improve the perception stack for real-time performance and safety.

Skills

NATO citizenship

Rust

C

C++

Python

Embedded hardware

Cross-functional collaboration

Technical communication

Education

MSc in Electrical Engineering / Computer Engineering

Job description

The Vydar Group in Delft is seeking a Machine Vision Engineer to design, implement, and optimize vision algorithms for constrained embedded hardware, focusing on NPU accelerators supporting autonomous defense platforms.

You will train detection, tracking and segmentation models, quantize and compile networks, profile latency and power, and collaborate with software, AI, electrical and mechanical engineering teams to deliver real-time, safe, and reliable perception stacks.