RL Evaluation Engineer: Build Production Benchmarks

Invisible, Inc.; Northern, New York (KY, NY); Hybrid; USD 150,000 - 210,000; Full time

Benefits offered by this job

Bonuses

Equity

Job summary

Invisible Technologies is building a reinforcement learning capability within its research organization. As a Research Engineer, you will design evaluation methods, build scoring frameworks, and ship production systems that measure frontier models for enterprise clients.

You’ll influence methodology, not just implementation, and collaborate across ML, solutions, and research teams to drive impact. This role emphasizes turning ambiguous research questions into working systems with

Qualifications

• Production-quality code written daily; this is a hard requirement.
• Fluency in Python and comfort across the modern ML stack.
• Experience building evaluation systems, RL environments, or training and inference infrastructure.
• Hands-on experience with modern agentic flows.
• Familiarity with reinforcement learning methods and frontier evaluation.

Responsibilities

• Design benchmarks and RL environments to measure real model capability.
• Originate evaluation methodology and translate into scoring frameworks and architectures.
• Write and ship production code implementing designs.
• Build and maintain data pipelines for evaluation runs.
• Run analyses to verify results and ensure reproducibility.
• Collaborate with Research Scientists, Solutions Architects, and ML engineers.

Skills

Python

ML Stack

Reinforcement Learning

Agentic Systems

Research to Production

Tools

Git

Docker

Kubernetes

Job description

Invisible Technologies is building a reinforcement learning capability within its research organization. As a Research Engineer, you will design evaluation methods, build scoring frameworks, and ship production systems that measure frontier models for enterprise clients.

You’ll influence methodology, not just implementation, and collaborate across ML, solutions, and research teams to drive impact. This role emphasizes turning ambiguous research questions into working systems with