R&D Deep Learning Intern

Job description

About

Kinetix is a leading Generative AI startup developing Artificial Intelligence technologies in the field of 3D animation.

Our goal is to create technologies that make 3D animation creation accessible to consumers (e.g., in video games) and more efficient than ever for professionals. Our current innovations include AI-powered Video to Animation and Text to Animation generation technologies.

Kinetix raised $12 million in Seed Funding with US-based fund Top Harvest Capital and with participation from major executives from the gaming industry. This investment supports the development of our AI models as well as our deployment in video games and 3D content platforms.

Job Description

The long-term goal of Kinetix's R&D team is to build, train and deploy a computer vision model that extracts 3D animation of a humanoid character from a monocular video. The quality of such models can be evaluated quantitatively (through metrics), but this does not capture the "pleasantness" or "realism" of the motion, which has to be evaluated qualitatively by a human.

The main objective of this internship is to develop a pipeline allowing for an easier and more objective comparison of the outputs of human pose estimation models. It consists in :

· 
Understanding the key aspects that matter when evaluating the quality of a human pose estimation model. Based on those key aspects, aggregating videos to build a relevant test set which will allow a human expert to assess motion realism.

· 
Designing, building and maintaining a human interface to allow for comparison and evaluation of different ML models on the task of human pose estimation

· 
Exploring state-of-the-art human pose estimation models, and building them in order to evaluate their outputs

· 
Producing a comprehensive classification of the methods used in state-of-the-art academic papers, as well as listing their strengths and weaknesses

Preferred Experience

You have the ideal profile if you :

· 
are passionate about technical deep learning problems

· 
want to develop and experiment with various state-of-the-art methods

· 
are autonomous and want to work on a task with significant ownership

· 
have some experience in Python, development and machine learning libraries

· 
are curious about 3d animation, and have a desire to learn and progress

Recruitment Process

· 
The first round is a presentation and human fit interview with the R&D Lead

· 
The second round is a technical interview with one or two R&D team members

Additional Information

·  Contract Type: Internship (Between 4 and 6 months)
·  Location: Paris
·  Education Level: Master's Degree
·  Experience: < 6 months
·  Possible partial remote