- Stronghold AI
- Paris, IDF
- Full-Time
- 12 days ago
Senior AI Researcher.
Before you go
Before you leave us, sign up for our email alerts
We don't do job spam, just the best digital jobs delivered straight to your inbox.
Senior AI Researcher: our view in 3 lines...
- The Role:This role is for a senior researcher who develops machine learning models that detect and track objects using video, audio and radio signals on small edge hardware.
- The Person:The person will develop and optimise detection, tracking and behaviour models, fuse detections from cameras, microphones and RF into one track, deploy models on edge GPU hardware, and feed field test results back into training.
- Requirements:The ideal candidate has machine learning experience, models shipped on edge or embedded systems, strong PyTorch and Python skills, and proven work in detection, tracking or classification of video, audio or RF signals.
About the role
Paris (preferred) · Senior · Full-time
About Stronghold AI
Stronghold AI is a European dual-use deep-tech start-up headquartered in Paris, building the AI platform for autonomous perception and cognition. Any sensor, any effector, one AI platform: our software turns cameras, microphones, RF and radar into decisions at the edge. AURA and LUMINA are the first two products built on it. We start with counter-drone because it is today's most urgent need; our ambition is to power perception and cognition across autonomous systems. Our mission is sharper awareness, faster decisions, and ultimately to protect lives.
The role
Our models run on small computers in the field, not in data centres. They have to find a drone in noisy video, audio and RF signals, in seconds, with whatever data the day gives them. As our senior AI researcher, you will own that problem end to end: read the literature, train on real recorded data, deploy to the edge, and watch it run in field tests. Where existing methods fall short, you will have the freedom to invent new ones. You do not need to arrive as a drone expert. A strong machine-learning background and the will to learn the domain alongside our RF and embedded engineers is enough.
What you will do
New approaches: create new methods where off-the-shelf models fall short, and see them run in the field
Models: develop and optimise models for detection, tracking and behaviour understanding
Fusion: combine detections from cameras, microphones and RF into one reliable track
Edge deployment: deploy models on edge GPU hardware
Data strategy: decide what we record in field campaigns, and how it becomes training data
Evaluation: design tests that predict field performance, not just benchmark scores
Robustness: harden models for cluttered, adversarial and low signal-to-noise conditions
Field loop: feed results from field tests back into training
Integration: work with embedded, RF and software engineers to put capability into deployed systems
What you will bring
5+ years of applied machine learning (or a PhD plus 3 years), with models shipped on edge or embedded systems
Strong PyTorch, and hands-on optimisation for speed, size and power (quantisation, pruning, distillation)
Proven work in detection, tracking or classification of video, audio or RF signals in difficult conditions
You read papers, reimplement what matters, and can tell when a result won't survive real data
Strong Python and solid engineering habits: testing, version control, reproducible experiments
Clear communication, including design notes and explaining trade-offs
Working English; French is a plus
Don't tick every box? Apply anyway. Ability and impact don't fit neatly into a list.
Nice to have
Edge deployment tools such as ONNX, TensorRT or CUDA
Multi-sensor fusion (vision, infrared, RF, audio)
RF signal classification (modulation recognition, emitter identification)
Audio or acoustic event detection
Self-supervised learning or anomaly detection on sensor data
Active learning, synthetic data or simulation
Experiment tracking (MLflow, Weights & Biases or similar)
C++, Rust or Go
Publications in top venues (valued, not required)
Who you are
A researcher who ships. You care about what runs in the field, not only what scores well on a benchmark, and you are happy to learn whatever the problem needs.
How we work
You judge your work by how it performs in the field, not on the bench.
You'll back an idea before it's proven, and you're honest enough to drop it if the evidence says no.
You raise problems early, own your work through to delivery, and help new colleagues get up to speed.
You design for cost, robustness and the user's conditions from day one.
You want your work to matter, and you hold yourself to a high standard of honesty about results.
What we offer
Early stock options: a real share in what we build, at the stage where it counts most
A ground-floor seat: you join while the platform is taking shape, and your work decides what it becomes
Your work in the field within weeks, not years
A small, senior team where you own problems end to end
Room to grow as new products and teams are built on the platform
Work that matters: protecting people, and the security of Europe and its allies
Equal opportunities
We are an equal opportunities employer and welcome applications from all qualified candidates. Because of export control and security rules, this role may require eligibility for security clearance. We discuss any restrictions early in the process.
How to apply
Email [email protected] with your CV and a short note on relevant work. Links to research, code or systems you have built are welcome.
