Tl;dr
Come and build LLM pipelines and agents, that help the largest companies modernise their legacy codebases.
Background
All large companies are stuck with legacy code running critical systems, written in languages that are dying and running on architectures that are increasingly expensive to maintain.
Our code translation pipeline automates the menial task of converting code from one language to another.
Previously human engineers would have had to read the legacy code and translate the codebase by hand, file by file, into the target language.
We're pioneering a novel approach to this problem, and leveraging LLMs to produce readable code in a fraction of the time.
What’s the role?
AI Agents are currently the bleeding edge of applied AI research, and you may have seen projects like Devin, OpenDevin, SWE-Agent, or GitHub Copilot Workspace.
This role is to build an AI Agent to perform refactoring of the transpiled code, that could only be possible with an LLM.
You’ll need to collaborate with a small but talent-dense team to design and implement the AI Agent.
You'll love what we're working on if you're interested in:
LLM Agents
New LLM models and architectures
Data curation, generation, and even some fine-tuning
Programming languages and compilers
You'll be joining a small group of engineers who are obsessed with the challenges posed by code translation, and who want to build tools to improve the process of writing and maintaining software.
Key requirements
Advanced knowledge of Python
Hands on experience with the state of the art LLMs
Strong fundamental understanding of LLM pipelines and agents
Also nice to have (bonus points)
Experience evaluating LLM pipelines
Experience fine-tuning models (e.g. PyTorch/Axolotl)
Understanding of programming language design
Rust experience
Benefits and Perks