Artificial Intelligence / Machine Learning Engineer

Come change the face of robotics with us

Created in 2021, Enchanted Tools is a team of 50+ mechatronics and AI experts united to create a groundbreaking new species of service robots, the Mirokaï. Performing critical social logistics tasks in public environments, the Mirokaï address worker shortage in healthcare (hospitals, nursing homes...) and other essential sectors (hospitality, industry...). Enchanted Tools is led by Jérôme Monceaux, former EVP at a leading robotics company.

In November 2022, we presented our prototype, developed in 12 months only.
We have also defined our identity and started working with talented and passionate people. Our company is growing quickly and we have a lot of exciting projects coming up. Now is a great time so join us !

👉 Click here for more information

We are looking for an Artificial Intelligence / Machine Learning engineer, for end-to-end interaction modelisation and integration on embedded systems, applied to social robotics.

Among the challenges encountered in establishing a natural interaction between a human and a robot in the "real world", one of the difficulties is to understand the orders given to the robot by a user but also to ensure that it only interacts when it is appropriate. Thus, the robot must be able to be in the same room as two humans who are having a discussion without reacting to each turn of speech between them, but only if one of them speaks directly to it. To do so, the robot needs to understand its environment well while remaining discreet without interacting "too much".

The purpose of this role will be to establish and evaluate different methods of multimodal data fusion applied to different problems of Enchanted Tools robots.
The applications will begin with the improvement of voice recognition thanks to audio/video fusion, the detection and recognition of the current speaker, and whether he is addressing the robot.

For this open position, we are looking for someone to help us :

• Participate in or enrich existing or new interaction corpora with our robots and analyze this data

• Set up models for analyzing these corpora, both with traditional methods and with approaches based on deep learning. Models will always be based on two or more modalities. A focus will always be on the interpretation of the results obtained, if possible in natural language

• Improve the model efficiency with the generation of synthetic data and improvement of our simulation solutions to robustify our solutions 

• Optimize and accelerate the models inference to bring the best approaches on board the robot, for a synthesis of appropriate behaviors

• Carry out a state of the art of the existing solutions and an inventory of the problems of the robot which could be solved by these approaches.

In this position, you will be part of a team to improve our solutions of human-robot interaction into the embedded environment that is our robot

Keys responsibilities of this position:

• Be part of the design of the full interaction pipeline, from audio and video signals analyzing to behavior and speech generation

• Participate in the continuous improvements of our simulation solution as well as our control interface of the robot (like WoZ or audio based): give feedback in usage, explore how to bring up models in the virtual world as well as the real one and evaluate them

• Conception and development of tests, documentation and CI

• Participation in interactions' stack integration and improve variability in the design of animal like reactions to ease communication with human while expressing affective states

• Deployment and testing of the solution on an actual robot

We are looking for a Machine Learning expert with a Master or a PhD in Computer Science, Animation, Affective computing or Robotics, and > 3 years of relevant professional experience, showing initiative and decision making in the design of computer solution development

Required:

• Mastering of deep learning techniques (training, optimization, quantification, pruning of DNNs)

• Experience in Artificial Intelligence methods applied to Human-(Agent|Robot|Computer) interaction with interest in algorithms for Computer Vision, NLP, Body language recognition and generation (Pose, Gestures, Facial expression,…)

• Good programming skills: Python, C++, Rust and experience with AI frameworks such as PyTorch or Tensorflow

• Proficiency with software life cycle and good practices: Linux environment, Git workflow, CI/CD, unit testing

Appreciated:

• experience with embedded/on edge AI models

• knowledge of cognitive and/or social sciences (Psychology, Linguistics etc.)

• skills in MLOps (knowledge of MLflow or Databricks, Onnx, TensorRt)

• Synthetic data generation

• experience with 3D engines (Unity, Unreal Engine, Godot)

• experience with robotics simulation engines (MuJoCo, Gazebo, Isaac Sim...)

If you are motivated and ready to enchant the face of the robot world, then join us !