Back

Job offer - Intern AI engineer - LaVague March 24

Mithril Security is hiring!

About

Mithril Security is a deep tech cybersecurity & AI startup developing Confidential AI infrastructure and AI solutions.

Founded in 2021 by three friends, our initial focus has been to create a secure AI infrastructure enabling users to leverage AI with robust privacy guarantees. Our Confidential AI technology has gained recognition among key industry players. Notably, our work has been supported by the OpenAI Cybersecurity Grant Program, which funded our R&D efforts to deploy AI models securely in enclaves. Additionally, we are working with the Future of Life Institute on developing hardware-backed AI governance tooling. Mithril Security's innovative approach also caught the attention of the UC Berkeley SkyDeck accelerator, significantly accelerating our growth. Our team, now comprising ten dedicated professionals, has successfully raised $1.4M in Pre-Seed funding with 3 VC investors, including Polytechnique Venture.

Entering a new phase of our journey, we recently unveiled a new project called LaVague, focused on the broader goal of democratizing AI use to automate daily tasks. LaVague automates manual tasks by translating natural language instructions into seamless browser operations, simplifying what is often seen as mundane or time-consuming tasks. This project leverages our AI and cybersecurity expertise to ensure that these automation pipelines handling users' sensitive data remain private. We are currently seeking world-class AI interns to help scale this ambitious project.

Learn more about LaVAGUE here.

Job Description

Context

Aside from developing only cybersecurity components for others like OpenAI, Mithril Security is now extending its focus towards creating solutions aimed directly at end-users. 

LaVague is enhancing internet interaction by converting natural language instructions into browser actions, thereby automating tasks with high efficiency. It employs the latest technologies in LLMs (RAG, Chain of Thought, Few-Shot, etc.)  to develop a performant system to generate Selenium code to pilot a browser. This enables the automation of various tasks, from form completions to bill payments. 

The technology is built on open-source models and projects, which promote transparency and align with user interests. Furthermore, it supports local models that prioritize user privacy and control.

Consequently, we are searching for a world-class AI engineer intern to assist our team of AI engineers in constructing an AI product meant to have a huge impact. 

The Role

As an AI Engineer Intern, your primary task will be to enhance LaVague's performance to generate actions from users' queries and improve our overall backend infrastructure.

Key Responsibilities

AI engineering stack: Your primary focus will be developing a robust and scalable application built on top of an LLM. This requires software development knowledge 

Model Efficiency and Scalability: Implement strategies to increase the model's performance and scalability (RAG, few-shot learning, Chain of Thought, async requests to LLM server, automatic prompt optimization, etc.)

Performance Benchmarking: Regularly conduct performance benchmarking against key metrics to measure improvements. This involves setting up a robust framework for continuous testing and evaluation of the model.

Documentation and Reporting: Thoroughly document the development process, including the optimization techniques, results of various experiments, and insights gained. Prepare detailed reports and presentations to communicate findings and progress to the team.

Open Source Contribution and Engagement: Actively contribute to the development and release of features for this open-source project while also engaging with the online community in collaboration with the Community Team. This role involves ensuring that our developments are communicated transparently and contribute positively to the broader ecosystem

Stay Informed on AI Developments: Follow the latest trends and advancements in AI and NLP, especially regarding LLMs, to ensure LaVague incorporates the most effective and ethical practices.

Preferred Experience

At Mithril Security, we value two things:

  • Technical excellence (aka you get complex stuff done)

  • Reliability (aka work given to you is done in a predictable manner)

In addition, these elements will be valuable for this application:

  • Enrollment in a Bachelor's or Master's program in Computer Science, Artificial Intelligence, Machine Learning, or a related field.

  • Demonstrable knowledge of machine learning, NLP, and AI.

  • Proficiency in programming (Python, etc.) and familiarity with Deep learning frameworks (TensorFlow, PyTorch, etc.).

  • Strong analytical and problem-solving skills, with keen attention to detail.

  • Excellent written and verbal communication skills, comfortable working in a collaborative team environment.

  • Any proof of outstanding skills in the above domain: price, awards, attendance to hackathon and competitions…

We prefer candidates willing to work in person with the dev. Team in France. 

Mithril Security considers that internships make sense only as a first step toward full-time employment.

Recruitment Process

We start by giving you a take-home assignment to evaluate your ability to learn new concepts fast, solve problems, and your autonomy.

You will have a given time window, usually one week, to give your assignment. 

Once we receive your assignment, you will have interviews with the founders of Mithril Security.

This process can be very fast if you are reactive!

For more information, do not hesitate to contact us at contact@mithrilsecurity.ioPlease complete this WTJ form and add your portfolio to the Cover Letter section so we can learn more about you. We will contact you shortly!

Additional Information

  • Contract Type: Internship (Between 6 and 8 months)
  • Location: Paris
  • Occasional remote authorized
Apply Now