If you have ever been annoyed that a loop is not vectorising and gone to look at the disassembly to find out why, we would like to hear from you.
AI ThinkLab is a small team working on problems where the answer is measurable. There is no ambiguity about whether the work was good: the frame rate went up or it did not.
What is different about working here
- Depth over breadth. You will get properly good at one hard thing rather than shallowly familiar with ten frameworks.
- Real hardware on your desk. Boards, probes, power meters and thermal chambers — not a cloud instance pretending to be an edge device.
- Variety of silicon. Different SoC every few months, which is the fastest way to learn what is fundamental and what is vendor-specific.
- Your name on the work. We encourage engineers to write up techniques publicly, within whatever the client NDA allows.
- Senior review from day one. Including for interns.
Open roles
Where we are hiring.
Senior Embedded AI Engineer
You read disassembly for fun, you know why a loop that looks vectorised is not, and you have put a model on a board that had no business running it. C/C++, Arm intrinsics, PyTorch or ONNX, and a profiler you actually trust.
Model Optimisation Researcher
Quantisation, pruning, distillation and graph surgery on real models with a real accuracy floor — and the judgement to tell us when a technique from last week's paper will not survive contact with an NPU compiler.
Associates & specialist collaborators
We keep a small network of named specialists — GPU kernels, ISP and sensor bring-up, RISC-V and FPGA, functional safety — brought onto engagements that need them. If that is your niche, we would like to know you before we need you.
Research internships
A structured six-month programme for final-year and recent ECE/CSE graduates. Real hardware, a supervisor who reviews your work properly, and a contribution that ships or gets written up by the end of it.
How to apply
Email careers@aitl.ai with your CV and — more usefully — a short note about something you made faster and how you knew it worked. A link to code, a blog post, a thesis chapter or a paragraph in the email body are all equally welcome. We read every one.
No open role that fits? Write anyway. We hire good engineers ahead of headcount more often than we advertise.