Machine Learning Engineer Internship, Kili
Training data platform for enterprise AI
Job no longer available
Training data platform for enterprise AI
21-100 employees
Job no longer available
21-100 employees
Company mission
To help enterprise customers turn raw data into training data so that they can scale AI applications on their internal challenges to solve their issues.
Role
Who you are
- Are in your last year of engineering school or M2 with a specialization in AI and machine learning
- Enjoy working as a team player and learning from others
- Possess good communication skills: you can explain your work and the technical choice you made, both orally and in writing
- Demonstrate great programming skills in Python, and experience working with machine learning / AI packages (OpenAI, HuggingFace Transformers, Scikit-learn, Pytorch or Tensorflow, Pandas …)
- Are curious, humble, organized, and a relentless doer
- Are confident in working in an international environment
- [Optional] Have already been confronted with some components of our backend stack (Typescript/Javascript, Node.js, GraphQL, Docker)
- [Optional] Are experienced in software engineering (git, CI, linting, packaging, …)
What the job involves
- As an intern, you will be at the heart of the patent drafting AI system: drawing figures following guidelines and best practices
- Patent agents generally use drawings provided by the inventors, which they annotate, and sometimes draw some themselves
- To that end, you will have to follow closely the state-of-the-art evolutions of Large Multimodal Models and Large Language Models, to understand the latest capabilities in terms of text-to-image and text-to-text processing, specifically on schematics generation and understanding
- You will focus on a few scientific domains, for which you will find appropriate knowledge bases to produce a deep understanding of the underlying invention and as a consequence increase the relevance of the descriptions
- You will also gather the data to fine-tune the models toward objectives relevant to the application and define a scalable evaluation protocol to ensure the algorithm actually increases the drawing speed of the end user
- Finally, given the interactive nature of the process, you will show a proof of concept of the drawing acceleration of the developed inside a simple writing interface
- You will do research and development work on your internship subject (60% of the time)
- You will do Python software development work to integrate your work into the product (20% of the time)
- You will communicate your technical work (20 % of the time):
- Publication in a relevant workshop or conference on the work done
- Blog posts on the Kili blog on selected ML topics
Application process
- If your application matches the profile, a first short interview will be scheduled with one of the team members
- Then a technical test will be sent to you to assess your coding skills and some machine learning knowledge
- If you pass the technical test, we will ask you to pick one article in the references of this internship, study it at home (~ 1h) and present it during a final technical interview with a member of the team (30 min); You will be asked questions about the article
Our take
According to digital business research, artificial intelligence is one of the primary areas of exploration for businesses looking to take advantage of growing tech developments. However, as Kili rightly emphasise, the efficacy and reliability of AI depends largely on the quality of data used to train it. Good data requires good labelling and the ability to store it for future use, and Kili is hoping to furnish enterprise businesses with precisely that.
The data labelling market is fiercely competitive, but the relatively young Kili has made impressive strides in their native France already. A roster of high profile clients including BNP Paribas, Safran, and Louis Vuitton demonstrates its ability to net quality enterprise custom across a breadth of markets using disparate data types. Plus, it has cemented Kili a strong reputation from which it can springboard its planned expansion to the US market.
Here is where things may get tricky. The United States takes 40% of the global AI market, and has attracted a correspondingly strong crowd of competing startups and enterprise businesses. Kili will benefit from a reputable track record, but will have to play fierce to get ahead in the market. Its 2024-launched Patent AI Co-pilot may go some way to differentiating it in the market, helping attorneys draft better patents and halving response time for office actions.
Steph
Company Specialist at Welcome to the Jungle
Insights
-16% employee growth in 12 months
Company
Funding (2 rounds)
Jul 2021
$25m
SERIES A
Jan 2021
$6.6m
EARLY VC
Total funding: $31.6m
Company benefits
- The opportunity to work in an international and friendly work environment
- The opportunity to build an expertise on THE topic that will be the future: AI
- The opportunity to join one of the most exciting startups of the French Tech as it is taking off
- Flexible work organization and access to remote work
- Afterworks every Thursday and 2 off-sites per year
Company HQ
Quinze-Vingts, Paris, France
Leadership
François-Xavier Leduc
(CEO & Co-Founder)Previously co-founded the car-as-a-service company Tripndrive, then went on to freelance for The Boston Consulting Group.
Edouard d'Archimbaud
(CTO & Co-Founder)Previously a Managing Partner at Creative Research Ltd., who then worked as a Data Scientist at Capgemini Consulting, and moved on to become Head of the Data and AI Lab at BNP Paribas.
Salary benchmarks
We don't have enough data yet to provide salary benchmarks for this role.
Submit your salary to help other candidates with crowdsourced salary estimates.
Share this job
View 1 more job at Kili