GenAI Research Scientist, Post-Training Team

Company Description

Established in late 2020 by a small group of machine learning researchers, MosaicML empowers companies to generate state-of-the-art AI models independently using their own data. MosaicML strongly believes that a company's AI models can hold equal value to any other crucial IP, and that high-quality AI models should be accessible to all. This stems from the company's commitment to reducing the cost of training top-notch models and our dedication to sharing our knowledge on how to achieve this with the world, thereby allowing everyone to innovate and create their own models.

Since becoming a part of Databricks in July 2023 under the name GenAI Team, we have been enthusiastically enabling our customers to tackle the world's most challenging problems by creating and managing the world's best data and AI platform. We thrive on solving technical challenges and strive to empower our customers with superior data and AI capabilities.

Job Description

As a Research Scientist on the GenAI Team at Databricks, you will be tasked with staying up-to-date with the latest advancements in deep learning and pushing the scientific frontier by devising techniques that surpass the existing state of the art. You will be working collaboratively with a team of researchers possessing diverse backgrounds and technical training. Above all, you will maintain a customer-centric approach: our goal is to enable our customers to succeed in training large models, and we incorporate our scientific expertise into our products to make this possible.

Specifically, in areas like fine-tuning, RLHF, and post-training, you will build upon our pre-trained models by introducing capacities for them to follow instructions, alter behaviours to effectively cater to the requirements of our users, and extend them to new capabilities like longer contexts, higher resolutions, tool usage, and so on. You will study the latest techniques in these areas, enhance them to meet our needs, and generalize them to address our customers' needs.

You might be a good fit if you...

• Are comfortable working with large-scale LLMs ranging from tens to hundreds of billions of parameters.
• Possess robust ML and RL software engineering and scientific research skills.
• Have produced innovative research related to fine-tuning LLMs or improving pre-trained LLMs.
• Have specialized expertise in areas like fine-tuning, RLHF, LLM tool-use, etc.
• Are passionate about enabling real users to access your work and, more broadly, democratizing access to modern AI technology.
• Are motivated to work on LLM research that, against the general trend in the field, will be disclosed to the public.
• Have excellent communication skills and a desire to work in a small, fast-paced team.

This role does not require a PhD. We are open to hiring candidates with bachelor's and master's degrees and to recent graduates. We are also open to hiring candidates currently employed as "research engineers" at other companies.

Your Responsibilities

• Staying current with the research literature and thinking beyond the state of the art to meet our users' needs.
• Developing and implementing methods to enhance and extend model capabilities, reliability, and safety.
• Rigorously assessing these methods, communicating your findings, and integrating useful methods into production.

Pay Range Transparency

Databricks is committed to fair and equal compensation practices. The pay range(s) for this role listed below represent base salary range for non-commissionable roles or on-target earnings for commissionable roles. Actual compensation packages are dependent on several unique factors to each candidate, including but not limited to job-related skills, depth of experience, relevant certifications and training, and specific work location. Given these factors, Databricks utilizes the full width of the range. The total compensation package for this position may also include eligibility for an annual performance bonus, equity, and the listed benefits. For more information regarding which range your location falls into, please visit our page here.