Research Engineer
Antimetal builds an autonomous system for production engineering, providing a continuously updated 'world model' of a company's software stack combined with specialized AI agents that diagnose, fix, prevent, and answer questions about production issues. It serves engineering teams at software companies who need to operate complex production systems without constant manual intervention.
Funding
Projects
About Antimetal
Antimetal is building the autonomous layer between engineering teams and their running production systems. At its core is a live world model, a continuous understanding of how a company's stack behaves, built by connecting to existing observability, infrastructure, and code tools. On top of this world model sits an army of specialized agents (Proactive Patrol, Reactive Triage, Intelligence World Model, and an Agent Builder platform) that investigate incidents, trace failures, identify root causes, propose fixes, open pull requests, and carry out operational workflows on behalf of engineering teams. Unlike traditional observability tools that surface data for humans to interpret, or coding agents that only generate code, Antimetal aims to continuously investigate, operate, and improve production itself, while routing changes through existing approval flows such as pull requests or Slack approvals. The company serves software engineering teams, is SOC 2, GDPR, and HIPAA compliant, and is built in NYC.
Skills
About the Role
You'll build the intelligent systems that power Antimetal. You'll prototype new approaches, run experiments, and own the path from research to production. You'll work closely with platform and product to shape agent capabilities and contribute to evaluation methodology. You'll tackle infrastructure observability, a hard domain to model, where telemetry is high-volume, noisy, and ephemeral, and ground truth is approximate. You'll help build AI agents that understand this complexity and can reason about what's happening, why, and how to fix it, including making changes to code and configuration.
Requirements
- 4+ years of experience in applied ML, research engineering, preferably at a company shipping production AI systems
- Production experience contributing to agentic/LLM systems, including multi-step reasoning, reinforcement learning, fine-tuning, and orchestration
- Proven experience bringing work from prototype to production, using data and experimentation to drive product and architectural decisions
- Strong on ML fundamentals: statistical modeling, probabilistic methods, time-series analysis, evaluation methodology
- Real world expertise in one area of applied ML: search, statistical modeling, NLP, etc.
- Experience constructing and running end-to-end evaluation pipelines with real world data
- Proficient in Python and Typescript, with experience using common ML libraries and data engineering tools
- Strong problem-solving skills, with a focus on creating highly maintainable, scalable code
- Comfortable with ambiguity and iterative development, prototyping, and adapting quickly to feedback
Responsibilities
- Run experiments across research areas, analyze results, validate what works, and take successful approaches to production
- Partner with platform on live and offline evaluation pipelines, benchmarks, and synthetic data generation
- Build tooling that lets the team measure progress and iterate with confidence
- Apply and develop techniques from best-in-class AI Agents, ML, and SRE research to the problem domain
- Experiment with new approaches to reasoning, retrieval, codebase mapping, and agent architectures
- Collaborate with platform and product to integrate capabilities and productionize prototypes into scalable and reliable services
Benefits
- Competitive salary with generous equity grants
- Fully covered health, dental, and vision
- Retirement benefits
- Unlimited PTO
- Dinner on late nights
- Monthly fitness stipend
- Equipment provided
- Citi Bike + train commute benefits
