Twenty-four single-page cheat-sheets on the Databricks platform — concepts, one small snippet, and one interactive diagram each. Built to skim in the last hour before the interview. Every fact grounded in the official docs and verified June 2026.
One platform fusing a lake's cheap open storage with a warehouse's reliability — refined Bronze → Silver → Gold.
02GovernanceThe unified governance layer — one permission model, one lineage graph, one audit trail across every workspace.
03Storage FormatParquet wrapped in a transaction log — real ACID, time travel, and schema control on cloud object storage.
04IngestionA streaming source that ingests new files as they land — no manual bookkeeping of what's already processed.
Define the tables you want; the framework builds the DAG, the compute, and the incremental updates. (formerly DLT)
06OrchestrationThe native lakehouse scheduler — wire tasks into a DAG, trigger on a clock or new data. (formerly Workflows)
10StreamingRun the same DataFrame code over an unbounded stream — exactly-once results via checkpoints.
Code → parallel work: a driver plans, executors run, and nothing happens until an action fires.
08PerformanceLet AQE re-plan at runtime, pick the right join, and stop reading data you don't need.
09Delta OptimizationCompaction, data skipping, and the clustering model that replaces partitioning + Z-order with no rewrite.
11SQL & BIServerless SQL warehouses on Photon — the vectorized C++ engine that accelerates work with no code change.
12ComputeAll-purpose, jobs, SQL warehouses, serverless — plus access modes, policies, autoscaling, spot, and Photon.
Track and register with MLflow, govern in Unity Catalog, deploy behind one autoscaling REST endpoint.
14AI / MLServerless similarity search that turns Delta tables into governed vector indexes — the retrieval layer for RAG.
15AI / BIAsk governed data questions in plain English — Genie writes and runs the SQL; dashboards share the result.
19GenAI / AgentsDeclaratively build domain agents on your data — describe the task, it generates the eval and tunes the agent.
20GenAI / AgentsThe code-first way to author, log, evaluate, deploy, and trace production agents — built around MLflow.
21GenAI / EvalMeasure GenAI/agent quality with LLM-as-judge scorers, eval datasets, human labeling, and production monitoring.
22MLOpsMLOps extended for GenAI: prompts, RAG, and agents through one develop → evaluate → deploy → monitor loop.
23MLOpsTrain, register, deploy, and monitor classic predictive models — DevOps + DataOps + ModelOps under one UC plane.
An open protocol to share live data and AI assets across orgs and clouds — no copying, no lock-in.
17OLTPA managed Postgres-compatible OLTP database built into Databricks — low-latency txns next to your lakehouse.
18DevOps / CI-CDInfrastructure-as-code for the lakehouse — describe resources in YAML and ship them dev → prod with one CLI.
24Databricks AppsAct as the app's service principal (shared data) or as the signed-in user via OBO (per-user data) — plus forwarded identity headers.