Databricks
About this role
Databricks customers can accumulate hundreds of millions of assets: tables, notebooks, dashboards, pipelines, SQL queries and ML models. Finding the right one is the problem this team owns, and the posting says 2026 is focused on search ranking, query understanding, evaluation frameworks and asset coverage. The work is machine learning applied to relevance: ranking and retrieval models, NLP pipelines for query understanding and rewriting, and the offline and online evaluation framework that proves a ranking change actually helped. Applying LLMs to search relevance is called out as a requirement rather than a bonus. The bar is high at 10+ years of building search relevance systems in production or high impact research, and contributions to well used open source projects are on the requirement list. Bengaluru.
Who this is for
Required by the posting: a BS or above, with an M.S. or PhD preferred, in Computer Science or a related field. 10+ years of experience developing search relevance systems at scale in production or in high impact research environments. Experience applying LLMs to search relevance. Experience in one or more of query understanding, NLP, text mining, recommendations, personalisation, discovery or conversational AI. A strong understanding of computer science fundamentals. Contributions to well used open source projects.
The real day to day: driving development and deployment of ML based search and discovery relevance models integrated with Databricks products; designing automated ML and NLP pipelines for data preprocessing, query understanding and rewriting, ranking and retrieval, and model evaluation, to enable rapid experimentation; collaborating with product managers and cross functional teams on the search and discovery roadmap; and building a robust framework for evaluating search ranking improvements both offline and online.
The team context, in the posting's own terms: this sits in the Applied AI team, and the stated 2026 priorities are enhancing search ranking, improving query understanding, building robust evaluations and growing asset coverage. Databricks customers can hold hundreds of millions of assets, so the scale of the ranking problem is the interesting part of the job.
Location: Bengaluru, India, in Databricks India Private Limited.
Honest fit guidance: this is a specialist role, not a generalist senior one. Ten years of backend or general ML work without search relevance specifically will not match, because the requirement names relevance systems, ranking, and evaluation metrics as the core skill. Equally, the open source contribution line is listed as a requirement rather than a preference, which is unusual and worth taking at face value. If you have built ranking systems and can point to public work, the fit is strong.
The real day to day: driving development and deployment of ML based search and discovery relevance models integrated with Databricks products; designing automated ML and NLP pipelines for data preprocessing, query understanding and rewriting, ranking and retrieval, and model evaluation, to enable rapid experimentation; collaborating with product managers and cross functional teams on the search and discovery roadmap; and building a robust framework for evaluating search ranking improvements both offline and online.
The team context, in the posting's own terms: this sits in the Applied AI team, and the stated 2026 priorities are enhancing search ranking, improving query understanding, building robust evaluations and growing asset coverage. Databricks customers can hold hundreds of millions of assets, so the scale of the ranking problem is the interesting part of the job.
Location: Bengaluru, India, in Databricks India Private Limited.
Honest fit guidance: this is a specialist role, not a generalist senior one. Ten years of backend or general ML work without search relevance specifically will not match, because the requirement names relevance systems, ranking, and evaluation metrics as the core skill. Equally, the open source contribution line is listed as a requirement rather than a preference, which is unusual and worth taking at face value. If you have built ranking systems and can point to public work, the fit is strong.
Apply on company site
Opens databricks.com, the employer's own application page. Applying is always free.