Research Engineer - AI Cybersecurity & Vulnerability Discovery

Google DeepMind; Mountain View (CA); On-site; USD 174,000 - 252,000; Full time

Benefits offered by this job

Equity

Benefits

Job summary

Google DeepMind is seeking a research-focused Software Engineer embedded across the organization to run large-scale tests, prototype implementations, and design new architectures while remaining connected to research through collaborations with universities and labs.

You will work on Gemini and CodeMender, applying ML and cybersecurity knowledge to build both innovative research prototypes and scalable infrastructure.

Qualifications

• Bachelor's degree in Computer Science, or a related technical field, or equivalent practical experience.
• 5 years of experience with large language models (LLM), including pre-training or fine-tuning.
• Experience with Cyber Security (in particular, application security and vulnerability research).
• Experience with running large-scale data processing pipelines.
• Experience collaborating on or leading applied research projects.

Responsibilities

• Contribute to Gemini Cyber, collaborating across organizations to build high-quality training recipes, cybersecurity datasets, benchmarks, and evaluation platforms for vulnerability finding and fixing.
• Build datasets (RL environments, SFT data) and iterate on post-training recipes.
• Integrate novel ML and agentic techniques into research prototypes.
• Build evaluation pipelines, benchmarks, and leaderboards to rigorously measure agent performance.
• Optimize system performance, latency, and resource utilization for serving large models.

Skills

LLM experience

Education

Bachelor's degree in Computer Science or related field

Job description

Google DeepMind is seeking a research-focused Software Engineer embedded across the organization to run large-scale tests, prototype implementations, and design new architectures while remaining connected to research through collaborations with universities and labs.

You will work on Gemini and CodeMender, applying ML and cybersecurity knowledge to build both innovative research prototypes and scalable infrastructure.