10/11/2024 updated
100 % available
Research and Software Engineer in ML, Computer Vision and Robotics
Fontainebleau, France
Worldwide
2 Master of Computer Science (Stanford University & Ecole Polytechnique)Artificial IntelligenceComputer VisionAutonomous SystemBig DataCloud ComputingNvidia CUDADatabasesExtract Transform Load (ETL)FrenchMachine LearningNatural Language ProcessingNumPyPrototypingRobotics Design and ProductionScientific Computating
Hi, I’m Josselin Somerville. I’ve had the opportunity to work at major tech companies like Google, NVIDIA, and Meta, as well as in startups like Zelros and Poligma, where I gained solid coding skills and a diverse range of experiences. I studied at two top universities—Stanford and École Polytechnique—which gave me a strong foundation in Computer Vision, Natural Language Processing, Reinforcement Learning, Data Science, and general Software Engineering.
Throughout my career, I've worked on various projects and learned a lot of programming languages and tools along the way. My biggest strength is my versatility: I’m quick to learn new skills and adapt to different challenges. I’ve even published some of my work in conferences like NeurIPS and ICRA.
I may not always know the exact tech stack you’re using, but I’m confident I can learn it quickly and get the job done. I’m fluent in both English and French, with dual citizenship in France and the UK. In my spare time, I enjoy working on robotics and prototyping—feel free to ask me about it!
You can visit my website: https://josselinsomervilleroberts.github.io or my LinkedIn: https://www.linkedin.com/in/josselin-somerville-roberts/
SKILLS:
Machine Learning and Computer Vision
Extensive experience in machine learning, computer vision, and AI, including work on perception software for autonomous vehicles and research in vision-language models.
Software Engineering
Proficient in multiple programming languages and frameworks, with experience in developing software for various applications including robotics, AI, and data processing.
Robotics
Expertise in robotics, including projects on climbing robots, robotic arms, and autonomous systems.
Data Processing and Analysis
Experience in data ingestion pipelines, ETL procedures, and working with large datasets for machine learning and computer vision tasks.
Deep Learning
Knowledge of deep learning techniques, including reinforcement learning and natural language processing, applied in various research and practical projects.
Full-Stack Development
Capability in full-stack development, including front-end technologies like Angular and React, and back-end systems using various databases and cloud technologies.
Scientific Computing
Proficiency in scientific computing tools and libraries such as NumPy, pandas, and CUDA for high-performance computing tasks.
Throughout my career, I've worked on various projects and learned a lot of programming languages and tools along the way. My biggest strength is my versatility: I’m quick to learn new skills and adapt to different challenges. I’ve even published some of my work in conferences like NeurIPS and ICRA.
I may not always know the exact tech stack you’re using, but I’m confident I can learn it quickly and get the job done. I’m fluent in both English and French, with dual citizenship in France and the UK. In my spare time, I enjoy working on robotics and prototyping—feel free to ask me about it!
You can visit my website: https://josselinsomervilleroberts.github.io or my LinkedIn: https://www.linkedin.com/in/josselin-somerville-roberts/
SKILLS:
Machine Learning and Computer Vision
Extensive experience in machine learning, computer vision, and AI, including work on perception software for autonomous vehicles and research in vision-language models.
Software Engineering
Proficient in multiple programming languages and frameworks, with experience in developing software for various applications including robotics, AI, and data processing.
Robotics
Expertise in robotics, including projects on climbing robots, robotic arms, and autonomous systems.
Data Processing and Analysis
Experience in data ingestion pipelines, ETL procedures, and working with large datasets for machine learning and computer vision tasks.
Deep Learning
Knowledge of deep learning techniques, including reinforcement learning and natural language processing, applied in various research and practical projects.
Full-Stack Development
Capability in full-stack development, including front-end technologies like Angular and React, and back-end systems using various databases and cloud technologies.
Scientific Computing
Proficiency in scientific computing tools and libraries such as NumPy, pandas, and CUDA for high-performance computing tasks.
Experience
Professional experience0 - 5 years
Freelancer since2024
Languages
EnglishNative speakerFrenchNative speaker
Project history
Worked on the DINO team, supervised by Piotr Bojanowski. Used World Models as a training objective for a self-supervised Video model using DINOv2 as the image encoder.
Supervised by Percy Liang. Collaborated on HELM (Holistic Evaluation of Language Models), a package to evaluate more than 80 LLMs on hundreds of scenarios. Implemented a Model Critique evaluation, interfaced many Human Evaluation APIs, and refactored the framework to simplify the user experience.
Developed new image augmentations to simulate objects on a camera lens using Perlin noise-based effects with several layers of Gaussian diffusion, noise addition, and various distortions. Improved the IOU score of a network that predicts camera occlusion.