Relari is defining the new category of data-driven development infrastructure for GenAI applications. We work with top AI teams from leading startups to large enterprises, and we are growing at a blazing-fast rate. Our founding team are experts on AI system production and we are well-funded by top-tier VCs.

As a founding research engineer at Relari, you will conduct cutting-edge research and solve the most critical challenges facing our society’s adoption of foundation models.

In this role, you will:

• Design and implement innovative features for a state-of-the-art Generative AI Testing and Simulation stack.
• Conduct novel research to push the boundaries in foundation model evaluation, model fine-tuning, synthetic data generation amongst other fields.
• Experiment with the latest findings in AI research and proactively incorporate best practices into production.
• Collaborate with top research labs to publish impactful papers.

You’re a good fit if:

• You have been a research scientist at a leading academic institution or top industry research group.
• You have published and reviewed papers in top academic conferences (NeurIPS, ICML, ACL, etc.).
• You thrive in a fast-moving environment and view unstructured environments as opportunities to identify the most impactful work and define the future success of the company.
• You are ownership-minded. You don’t wait for processes to get work done; instead, you proactively contribute wherever you can to drive success.

Bonus points if you have:

• Experience training and fine-tuning large foundation models.
• Shipped an AI product or feature.
• Contributed to open source Generative AI projects.

Our testing and simulation stack leverages a suite of foundation models and traditional machine learning techniques to generate large-scale, high-fidelity synthetic data for rigorous evaluation and systematic fine-tuning. There's a heavy research angle to our tech stack as we are on the cutting edge of defining the best practices in building reliable AI applications. We collaborate with top institutions like MIT to publish research papers and open source part of our stack (continuous-eval).