Research Engineer (multiple positions)

DepthFirst•San Francisco, CA

14d

About The Position

About DepthFirst AI We believe that software is the foundation of modern civilization - yet vulnerabilities threaten its integrity, security, and resilience. We are on a mission to solve security. DepthFirst is building intelligence to detect and remediate critical software vulnerabilities. We are training and scaling security AI agents to discover zero-days vulnerabilities, from large customer codebases to popular open source software. Our founding team includes expertise in security and LLMs (with technical leaders from DeepMind, Databricks, Square, and Faire). We are looking for strong technical people who are interested in working at the intersection of AI, Security and Infrastructure. About this role: We’re seeking an experienced Research Engineer to join our effort in building and training AI agents for vulnerability discovery and exploitation. We are building a technology capable of finding the next Log4J at scale, finding and remediating vulnerabilities in customer and open source codebases. We are looking for strong engineers with strong intuition, experience in model evaluation and benchmarks. Reinforcement Learning experience is a plus. Your work will play a crucial role in building a product that aims to redefine how companies do security.

Requirements

3+ years of full-time experience in AI research engineering in machine learning, deep learning, and/or natural language processing
Experience with developing machine learning models at scale from inception to business impact
Programming experience in Python and hands-on experience with frameworks such as PyTorch
A bachelor's degree in Computer Science/Software Engineering or equivalent industry experience
A love for technology, and an insatiable curiosity for new tools to tackle real problems
Capable of solving complex problems with simple solutions. Building reliable and scalable products, making right trade-offs along the way
A tendency to leave things in a better way than you found it

Nice To Haves

Reinforcement Learning experience is a plus.

Responsibilities

Build State-of-the-Art AI Agentic pipelines optimized for discovering intricate software vulnerabilities and generate exploits
Design, Maintain and Evolve evaluation benchmarks for AI agents that reflects the problems our users face
Design & Develop training procedures and reinforcement learning environments to train security coding agents
Work on a Product that Solves a Critical Problem - and we already have a handful of customers who have found it valuable in fixing some eye-opening vulnerabilities within the first few days of using our product.