Embedded Software Engineer, AI Accelerators
Aliensense (Masdar City, UAE)
Aliensense is the physical AI company. We build a modular compute and sensor platform that gives robots the perception, reasoning, and real-time control they need to operate in the physical world. Our hardware stacks NVIDIA tech with custom GMSL camera modules, CAN-FD buses, and a dedicated AI accelerator tier. We are based in Masdar City and backed by deep-tech investors across the GCC and Europe.
The Role
We are hiring an Embedded Software Engineer to own the AI inference layer on our robotic modular platform. You will integrate AI-accelerators (and optionally Jetson's DLA/GPU) into our perception stack, taking trained models from ML engineers and making them run fast, deterministically, and with bounded latency on our embedded platform.
This is a hands-on, end-to-end role: from flashing firmware and writing low-level runtime code to profiling inference on real hardware and handing off tested pipelines to robotics engineers.
This is a hands-on, end-to-end role: from flashing firmware and writing low-level runtime code to profiling inference on real hardware and handing off tested pipelines to robotics engineers.
What You will Do
- Own the model deployment pipeline: ONNX / TFLite / PyTorch export → Accelerator SDK / Palette compiler → runtime integration on robotic platform
- Write and maintain C/C++ inference wrappers and camera-to-accelerator data paths with deterministic latency budgets
- Profile and optimise models for the MLSoC AI-accelerator: layer fusion, quantization (INT8/FP16), memory layout, bandwidth bottlenecks
- Integrate accelerator outputs (detection, depth, segmentation, SLAM features) into our ROS 2 perception stack
- Collaborate with the firmware team to ensure camera triggers, GMSL frame timestamps, and inference timestamps are tightly correlated
- Maintain the inference runtime as a versioned, reproducible component of platform's software stack
- Write and maintain C/C++ inference wrappers and camera-to-accelerator data paths with deterministic latency budgets
- Profile and optimise models for the MLSoC AI-accelerator: layer fusion, quantization (INT8/FP16), memory layout, bandwidth bottlenecks
- Integrate accelerator outputs (detection, depth, segmentation, SLAM features) into our ROS 2 perception stack
- Collaborate with the firmware team to ensure camera triggers, GMSL frame timestamps, and inference timestamps are tightly correlated
- Maintain the inference runtime as a versioned, reproducible component of platform's software stack
Requirements
- 3+ years of embedded or edge ML engineering experience
- Hands-on experience deploying models to a dedicated AI accelerator (SiMa.ai, Hailo, Coral, Kneron, Horizon, or similar MLSoC — not just GPU/CPU)
- Strong C/C++ (modern C++17/20); Python for tooling and model export
- Understanding of quantisation-aware training, post-training quantisation, and the performance trade-offs
- Familiar with ONNX, TFLite, or equivalent model interchange formats
- Experience reading and interpreting profiler output (bandwidth, compute, memory) on constrained hardware
- Hands-on experience deploying models to a dedicated AI accelerator (SiMa.ai, Hailo, Coral, Kneron, Horizon, or similar MLSoC — not just GPU/CPU)
- Strong C/C++ (modern C++17/20); Python for tooling and model export
- Understanding of quantisation-aware training, post-training quantisation, and the performance trade-offs
- Familiar with ONNX, TFLite, or equivalent model interchange formats
- Experience reading and interpreting profiler output (bandwidth, compute, memory) on constrained hardware
Nice to Have
- Familiarity with NVIDIA Jetson (DLA, CUDA, TensorRT, Deepstream)
- GStreamer / V4L2 camera pipeline experience
- ROS 2 node development
- Background in SLAM, visual odometry, or 3D perception
- GStreamer / V4L2 camera pipeline experience
- ROS 2 node development
- Background in SLAM, visual odometry, or 3D perception
What We Offer
- Hands-on work with cutting-edge edge AI silicon
- End-to-end ownership — from silicon bringup to live robotics demos
- Small, senior team with no bureaucracy
- Competitive compensation
- Masdar City HQ — the UAE's deep-tech hub
- End-to-end ownership — from silicon bringup to live robotics demos
- Small, senior team with no bureaucracy
- Competitive compensation
- Masdar City HQ — the UAE's deep-tech hub
Apply
careers@aliensense.com · Subject: `AI Accelerators Engineer`