Company Description

Founded in late 2020 by a small group of machine learning researchers, MosaicML enables companies to create state-of-the-art AI models from scratch on their own data. From a business perspective, MosaicML is committed to the belief that a company’s AI models are just as valuable as any other core IP, and that high-quality AI models should be available to all. From a scientific perspective, MosaicML is committed to reducing the cost of training state-of-the-art models - and sharing our knowledge about how to do so with the world - to allow everyone to innovate and create models of their own.

Now part of Databricks since July 2023 as the GenAI Team, we are passionate about enabling our customers to solve the world's toughest problems by building and running the world's best data and AI platform. We leap at every opportunity to solve technical challenges, striving to empower our customers with the best data and AI capabilities.

Job Description

As a Research Scientist on the GenAI Team at Databricks, you will be responsible for keeping up with the latest developments in deep learning and advancing the scientific frontier by creating new techniques that go beyond the state of the art. You will work together on a collaborative team of researchers with diverse backgrounds and technical training. And most importantly, you will love our customers: our goal is to make our customers successful when they train large models, and we encode our scientific expertise into our products to make that possible.

Specific to model evaluation, you will work on designing high-performance, scientifically motivated evaluation suites to enable our research scientists and our customers to make critical decisions when training and deploying state of the art generative models for text, images, and other domains. You will get to push the state of the art in evaluation of tool use, code generation, RAG, safety and toxicity evaluation, and model-based evaluation. You will also help build and design systems to empower our customers to build the next generation of evaluations for their custom, domain-specific generative models..

You might be a good fit if you…

• Are comfortable working with large-scale LLMs in the 10s to 100s of billions of parameters.
• Have solid ML and text/data processing engineering and scientific research skills.
• Have deployed LLM systems to solve real world problems and have designed clever ways to evaluate system performance.
• Have worked on safety, jailbreaking, or red-teaming LLMs.
• Are passionate about getting your work into the hands of real users and - more broadly - democratizing access to modern AI technology.
• Are motivated by working on LLM research that - contrary to the general trend in the field - will be disclosed to the public.
• Have strong communication skills and a desire to work on open-ended problems over which you will have a large degree of autonomy.

A PhD is NOT required for this role. We are open to hiring candidates with bachelor's and master's degrees and to new graduates. We are open to hiring candidates who are currently in "research engineer" roles at other companies.

Your Responsibilities

• Keeping up to date with the research literature and thinking beyond the state of the art to address the needs of our users.
• Developing and implementing novel methods to evaluate the capabilities of generative models for text, images, and other domains.
• Rigorously testing these methods, communicating the results of your findings, and putting those that are useful into production.