About the RoleWe're a ~15-person engineering team - made up of Olympiad medalists and published researchers - building infrastructure that aligns AI to real-world workflows through reinforcement learning environments and post-training data. We're hiring
Research Engineers to own the synthetic data pipeline: transforming domain-specific workflows into scalable, high-quality training tasks for AI agents.
This is a high-ownership, low-bureaucracy role. You'll be working in genuinely unstructured problem spaces where the roadmap is yours to define. Visa sponsorship is available.
What You'll Do- Build and maintain the end-to-end synthetic data pipeline, converting domain-specific workflows into realistic, structured, and challenging training tasks for AI agents.
- Collaborate with subject-matter experts to generate synthetic tasks across professional and technical domains.
- Design synthetic task generation methods that produce diverse, realistic, and learnable outputs.
- Build tooling to mutate, validate, and iteratively improve synthetic tasks at scale.
- Analyze model and agent performance on synthetic tasks to understand what they teach and where they break down.
- Develop metrics to quantify synthetic task diversity, realism, learnability, and overall quality.
What We're Looking ForRequired:- 2-4 years of experience in software engineering, ML engineering, or AI research - with a track record of shipping data pipelines, ML infrastructure, or synthetic data systems.
- Hands-on experience applying synthetic data research methods to build end-to-end data generation pipelines for AI/ML applications.
- Proficiency in Python; comfortable working in Linux environments with containerization tools such as Docker.
- Demonstrated understanding of synthetic data quality criteria and evaluation metrics (diversity, realism, learnability) and their limitations - from production or research work.
- Experience designing, implementing, or maintaining evaluation frameworks, benchmarks, or testing environments for AI agents or large language models.
- Experience building automated systems to generate, validate, mutate, or process structured datasets at scale.
- Proven ability to independently own and deliver technical projects end-to-end with minimal predefined requirements.
Nice to Have:- Experience detecting edge cases, inconsistencies, or quality issues in synthetic or algorithmically generated datasets.
- Experience creating synthetic tasks, data, or evaluations across multiple distinct professional or technical domains.
- Familiarity with reinforcement learning training paradigms, agentic AI workflows, or LLM post-training pipelines.
You'll thrive here if you:- Reason from first principles about task design, scoring, and failure modes.
- Are detail-oriented and naturally spot subtle inconsistencies in data and systems.
- Are energised by early-stage, ambiguous environments rather than frustrated by them.
- Communicate clearly and collaborate effectively across time zones.
Compensation & Benefits- Salary: $150,000 - $250,000 USD annually
- Visa sponsorship available
- Equity participation (early-stage startup)
LocationThis role is
on-site in San Francisco, CA. Candidates based in or willing to relocate to San Francisco are strongly preferred. The team also has a presence in Singapore.
