Staff Software Engineer | Search Quality

Databricks is the data and AI platform company behind the lakehouse architecture, and this is a staff engineering role on search quality. Search quality work sits between information retrieval and systems engineering: ranking and relevance, evaluation methodology, and the infrastructure to measure whether a change actually improved results rather than just moved them. In a platform as large as this the search surface spans documentation, data assets, notebooks and catalogue objects, so the problem is enterprise search over heterogeneous content rather than web search. The posting publishes no years figure anywhere, so we have recorded it as blank rather than guess, and it will not show under any experience filter here. Good fit if you want relevance engineering with real scale behind it.

Databricks publishes no experience figure anywhere in this posting. We have left the field blank rather than invent a number, so this listing appears in no experience band on the page. Treat the Staff title as the seniority signal: at Databricks that generally means deep technical ownership, the ability to set direction for a problem area, and influence beyond a single team. Ask for the expected band in the first call. What search quality means in this context: relevance and ranking work over the platform's own content surfaces, which for a lakehouse platform typically includes data assets, tables and catalogue objects, notebooks, documentation and internal search across a customer's workspace. The engineering split is usually part information retrieval (ranking models, query understanding, embeddings and hybrid retrieval) and part systems (indexing pipelines, latency, freshness, and the evaluation infrastructure that tells you whether a ranking change helped). The evaluation half is often the harder and more valuable one, because enterprise search has no equivalent of web scale click logs to lean on. Location is Bengaluru. Employer context that is genuinely relevant to fit: Databricks runs a very large India engineering presence and its Bengaluru requisitions today range from roles asking five years to one asking fifteen, so the seniority mix on this board is wide and the absence of a figure on this posting is not necessarily a signal that the bar is low. Good fit for an engineer with strong information retrieval instincts who wants to work on relevance where the corpus is structured enterprise data rather than open web text, and who is comfortable that a large part of the job is proving improvement rather than shipping features.