Research Engineer
About this role
Invisible Technologies needs a Research Engineer to solve a core problem in enterprise artificial intelligence: how to accurately and reproducibly measure what frontier models can actually do. The research organization is building out a dedicated reinforcement learning capability focused on evaluation methodology, benchmarks, and environments for top labs and enterprise clients.
In this role, team members take ambiguous research questions about model capabilities and turn them into functional scoring frameworks, evaluation architectures, and production systems. The work requires designing the methodology and writing the code to prove it out rather than just handing off specifications. Ideal candidates are self-directed engineers who enjoy translating complex research concepts into production-grade systems while partnering closely with research scientists, solutions architects, and machine learning engineers.
What you will do
- Design benchmarks and reinforcement learning environments to measure real model capability for enterprise clients and frontier labs.
- Originate evaluation methodology and translate it into scoring frameworks, rubrics, and evaluation architectures.
- Write and ship production code that implements your designs.
- Build and maintain the data pipelines that feed evaluation runs.
- Run analyses to establish whether results hold and ensure evaluations are reproducible.
- Partner with Research Scientists on methodology review, Solutions Architects on client requirements, and ML software engineers on underlying platform infrastructure.
What you need
- Write production-quality code daily as a hard requirement for the role.
- Demonstrate fluency in Python and comfort across the modern machine learning stack.
- Bring real experience building evaluation systems, reinforcement learning environments, or training and inference infrastructure.
- Possess hands-on experience with modern agentic flows.
- Show familiarity with reinforcement learning methods and how frontier models are evaluated.
- Have a proven track record of turning ambiguous research questions into working systems and publishing or shipping the result.
Benefits
- Bonuses and equity included in all full-time offers
Skills and tech stack
Questions candidates ask
What is the work location policy for this role?
The position is hybrid, with options available in New York and the San Francisco Bay Area. Invisible operates as a hybrid organization where most positions involve in-office collaboration tied to specific locations.
How is compensation determined?
Compensation includes base pay along with bonuses and equity in all full-time offers. Final pay is determined by factors including location, job-related experience, skills, knowledge, internal pay equity, and market conditions.
What kind of technical background is expected?
Candidates must write production-quality code daily in Python and feel comfortable across the modern machine learning stack. Real experience building evaluation systems, reinforcement learning environments, or training and inference infrastructure is required.
The application is on the official Invisible Technologies careers page.
Apply on company site