Research Engineer: Frontier AI Environments & Verification

Collinear AI, Inc.; Sunnyvale (CA); On-site; USD 160,000 - 400,000; Full time

Benefits offered by this job

Frontier projects

Equity

Direct impact

Competitive salary

Job summary

Collinear is seeking a highly capable MTS - Research to build data engines for frontier AI. You will craft ultra-realistic SimLabs and evaluation stacks used by leading AI labs to stress-test advanced agents.

The role spans computer use, enterprise tools, and coding with emphasis on reliability and safety. You will collaborate with founders and researchers, tackle RLHF/RLAIF domains, and contribute to scalable data pipelines and long-horizon agentic environments.

Qualifications

• Proficient in Python development and CLI-first workflows.
• Strong foundation in ML methods, model evaluation, and reproducibility.
• Experience conducting empirical research and producing data-driven results.
• Proven ability to work across domains with researchers and engineers.

Responsibilities

• Design and implement ultra-realistic, long-horizon simulation environments (SimLabs).
• Develop rigorous policy-aware verifiers and evaluations beyond benchmarks.
• Plan and execute post-training runs (CPT, SFT, RL) to demonstrate frontier performance.
• Collaborate daily with founders and researchers to shape the roadmap and advance AI reliability.
• Analyze model failure modes and create data pipelines that scale with test-time compute.

Skills

Python

ML foundations

Research experience

Reinforcement Learning

Education

Bachelor’s/Master’s/PhD in CS/Math/Physics or related field

Job description

Collinear is seeking a highly capable MTS - Research to build data engines for frontier AI. You will craft ultra-realistic SimLabs and evaluation stacks used by leading AI labs to stress-test advanced agents.

The role spans computer use, enterprise tools, and coding with emphasis on reliability and safety. You will collaborate with founders and researchers, tackle RLHF/RLAIF domains, and contribute to scalable data pipelines and long-horizon agentic environments.