AI Research Engineer

New York. Posted 2 months ago

The role

You will build and scale the infrastructure used by the research team to test and deploy agent behavior. You will design experiments, build evaluation pipelines, and turn research prototypes into production-ready systems.

What you will do

Build and scale research infrastructure including experiment harnesses and evaluation pipelines
Take research prototypes and make them fast, reliable, and production-ready
Partner with researchers to design experiments and frame hypotheses
Work with AI engineers, infrastructure teams, and product leads to bring research into production
Track developments in LLMs, agent architectures, and evaluation methodology
Prototype and evaluate prompting strategies, reasoning workflows, and tool-use policies

Required qualifications

5+ years of software engineering experience with a focus on backend development and scaled systems
Hands-on experimentation experience running, evaluating, and iterating on experiments
Ability to lead in fast-paced startup environments with limited resources and minimal structure
Strong experience with Python and web frameworks such as FastAPI
Experience deploying applications on AWS, working with ECS or Kubernetes, Postgres, and S3
Familiarity with observability and monitoring tools

Nice to have

A research engineer background, or prior experience as the engineer embedded in a research team
Experience building or operating agentic AI systems
Background in large-scale, complex, data-driven applications
Familiarity with AI or LLM-powered products, and with evaluation or benchmarking tooling

Benefits

Startup equity
Health insurance
Fertility benefits
Great tech setup stipend
Flexible time off
In-office snacks
Team happy hours and outings
Annual company offsite

About the company

Traversal is the AI Site Reliability Engineer (AI SRE) for the enterprise. The company builds systems that autonomously understand and reason across complex production environments to diagnose, fix, and prevent incidents.