# Databricks consulting in San Francisco, CA > The Bay Area's software industry, from SoMa's AI labs down the 101 to San Jose, needs no introduction, and much of its AI work is now maintenance. If you built a system quickly and now have to make it durable, governed and affordable, that's the work we do here. Canonical: https://www.techfabric.com/databricks-consulting-san-francisco Serving: San Francisco, Oakland, San Jose, Palo Alto, Berkeley, Mountain View Headquarters: 1530 E Williams Field Rd, Ste 201, Gilbert, AZ 85295. No office in San Francisco. --- ## Why San Francisco work is our work Your Snowflake bill, a vector database and a model spend nobody forecast all landed in one quarter, and finance wants to know which to cut. We consolidate onto Databricks with cost policies set during the migration and agents that inherit Unity Catalog grants, so the bill can be forecast and an auditor can see who read what. A SoMa AI startup usually hits this after its first enterprise deal, and a Mission Bay biotech hits it with its research data. ## Work a buyer here would recognise Canvass is a production Cloudflare application for account research and campaign review. Campaign preparation, approval and delivery states are kept distinct, and per-tenant isolation is enforced in the data layer, which is the study to read if your own system now has to show who approved what. ## The bill arrives before the governance does A lot of San Francisco's AI systems went from demo to production inside a quarter, on a per-token model API and a vector database nobody budgeted. The bill shows up first. Then an enterprise customer's procurement team wants an audit trail, and from January 2027 the California Privacy Protection Agency's 2025 rules on automated decision-making technology will give consumers notice and opt-out rights when such a system makes a significant decision about them. ## Questions ### How would a San Francisco data broker handle DROP deletion requests? As a scheduled pipeline that keeps a record of every request. From August 1, 2026, registered data brokers must access the Delete Request and Opt-Out Platform at least once every 45 days and process the deletion requests waiting there, according to the California Privacy Protection Agency. We'd pull each batch into a Delta table, match it against your identity data, and delete the matching rows from every downstream table that Unity Catalog lineage shows was built from them. Delta keeps old files for time travel, so the job also purges deleted rows from storage once the retention window passes, and we'd set that window short enough to finish well inside the 45 days. Sources: [the California Privacy Protection Agency](https://cppa.ca.gov/data_brokers/) ### How would Databricks handle ledger reconciliation for a San Francisco fintech? With a nightly match and a table of exceptions. We'd land processor settlement files and your internal ledger in Delta with change data feed switched on, match them every night, and park each break where operations can work it. Financial activities accounted for 119,300 jobs in the San Francisco metro in September 2026 (preliminary), says the Bureau of Labor Statistics. Time travel only reaches back as far as a table's retention setting, so each night's results would go into a snapshot table an auditor can query a year later. Unity Catalog's audit logs record who read card or account fields. Sources: [the Bureau of Labor Statistics](https://www.bls.gov/eag/eag.ca_sanfrancisco_msa.htm) ### How would a Bay Area utility put wildfire-season sensor data on Databricks? As a stream, joined to weather and asset records before anyone reads it. PG&E, based in Oakland, serves about 16 million people and runs 106,681 circuit miles of distribution lines, which gives a sense of how much line data a San Francisco Bay Area utility can be looking at. We'd stream sensor readings into Delta, join them to weather and asset records in a pipeline governed by Unity Catalog, and publish a risk table field crews can read on a tablet. The model scoring that risk is registered with a snapshot of the data it used, so after an event you can show exactly what it knew. Sources: [PG&E](https://www.pge.com/en/about/company-information/company-profile.html) ### We built this fast and now the model bill is a problem. That is the most common brief we get here. Model spend, retrieval infrastructure and the human review a system still needs get instrumented, and then the architecture questions become answerable. Usually the fix is durability and caching rather than a different model. ## What databricks consulting covers ### Migration off Snowflake, Synapse and Teradata Run through TechFabric Airlift, with executable playbooks for twenty-seven source platforms, signed migration certificates, and a cutover you can reverse at two in the morning. ### Unity Catalog governance Catalogue and schema design, a grant model expressed through groups, row filters and column masks, and lineage that survives a refactor. The boundary your agents and Power BI inherit. ### Lakeflow pipelines Ingestion and Lakeflow jobs with schema evolution handled rather than hoped for, medallion boundaries drawn on purpose, and jobs that recover instead of restarting from zero. ### Databricks Apps and agents Applications that run inside the workspace under their own service principal, and agents on Unity Gateway that inherit catalogue permissions instead of routing around them. ### Genie and AI/BI Metric definitions agreed with the person who owns the number, and a benchmark suite Genie has to keep passing, so the answer holds up when finance checks it. ### Cost that somebody can defend Cluster policies, warehouse sizing, file layout and the queries that actually pay. Instrumented at build time rather than discovered on an invoice. ## Working with us from Arizona Arizona does not observe daylight saving, so from March to November we are on the same clock as you and from November to March we are an hour ahead. Either way the working day overlaps almost completely. Most delivery is remote or hybrid. We travel for discovery, workshops and the opening weeks of a build.