As a Machine Learning Engineer, you will develop new AI models for our platform, trained on our 3D and video data.
You will start from large pre-trained models and adapt them to our data, generate the training data they need, and reduce them so that they run in real time on the hardware used in the field.
Every result must come with a confidence level that can be trusted in the field, and measuring it is part of the role.
You will work with our engineers in machine learning, computer vision, 3D graphics and mobile software.
We are looking for someone who has already delivered software in production, pays attention to detail and likes working on problems that are not yet solved.
Responsibilities
- Fine-tune large pre-trained vision and video models on our 3D and video data, including parameter-efficient methods such as LoRA
- Design and train new models for 3D and video understanding
- Own evaluation and calibration, so that every result carries a confidence that can be trusted in the field
- Distil and compress models to run in real time on field hardware: quantisation, ONNX/TensorRT
- Generate synthetic training data from our 3D reconstructions and reduce the gap with real data
- Build the training and evaluation pipelines, including pipelines that run on-premise
- Follow the research, build prototypes, and take the ones that work through to the product: specification, integration, release
- Ensure product quality through rigorous code standards, tests and CI/CD methodologies
- Stay current with advancements in vision foundation models, generative models, 3D vision and edge AI
Requirements
- Master's or PhD in Computer Science, Machine Learning, Applied Mathematics or related field
- At least 4 years of experience in machine learning or computer vision, including at least 2 years building products in industry
- Strong Python skills, and good C++
- Solid deep-learning foundations and hands-on experience with modern architectures: vision transformers (ViT), vision-language models (VLM), diffusion or video models
- Generative modelling and uncertainty estimation
- Experience fine-tuning large pre-trained models and making them smaller and faster
- Experience shipping software products, with a customer-oriented mindset
- The habit of writing tests and working in CI on a shared codebase
- Working proficiency in English; French is a plus
Nice to Have
- Research experience in the field (3D vision, generative models); publications or open-source work are a plus
- Synthetic data generation, simulation and sim-to-real transfer
- Deploying models on embedded GPUs or other edge hardware
- Aerial or satellite imagery and geospatial data: raster data, coordinate frames, georeferencing
- Background in robotics and unmanned autonomous systems
- Computer graphics and 3D: real-time rendering, 3D representations (meshes, point clouds, Gaussian splatting), GPU programming (CUDA, Vulkan, Metal)
- RTT (Reduction of Working Time)
- Bonuses and Stock Options
- Startup environment with hybrid work flexibility
Recruitment Process
- 01 Initial screening and interview with the CTO
- 02 Technical interviews, with senior tech team members
- 03 Final interview with the CEO
- 04 Quick decision-making and hiring process