AI Engineer
← All organizations

Enterprise data, analytics and AI

Databricks

Databricks provides a unified platform for enterprises to process data, run analytics and build AI applications. Data engineers use Lakeflow for batch and streaming pipelines, while Lakehouse provides data warehousing and Unity Catalog governs data and AI assets. Business users can ask Genie questions in natural language and receive summaries, tables and visualizations. Developers use Agent Bricks to build agents grounded in company data and Lakebase, its serverless Postgres database, to run transactional applications alongside analytics and AI workloads.

Founded in 2013, Databricks brought together Ali Ghodsi, Ion Stoica, Matei Zaharia, Patrick Wendell, Reynold Xin, Andy Konwinski and Arsalan Tavakoli-Shiraji. Ghodsi is CEO and Zaharia is CTO. Zaharia started Apache Spark at UC Berkeley in 2009, establishing the company's roots in distributed data processing. Its lakehouse research proposed combining warehouse analytics with machine learning over directly accessible open data formats. Delta Lake supplies ACID transactions and data versioning for data lakes, while MLflow manages machine learning workflows.

Databricks sells a managed commercial platform with consumption-based pricing measured in Databricks Units. In August 2026, the company reported that more than 20,000 organizations, including 70% of the Fortune 500, used its platform, and that its annualized revenue run rate exceeded $7 billion.

www.databricks.com ↗

3 talks

Newest first

2 speakers at AIE

Affiliations reflect their AIE appearances, not necessarily current employment.

Sandipan BhaumikData and AI technical leadOmar KhattabAssistant professor of electrical engineering and computer science

Start here

  1. On Engineering AI Systems that Endure The Bitter Lesson

    Start here to understand how Khattab interprets the Bitter Lesson and where he sees a continuing role for software engineering.

    Omar KhattabAI Engineer World's Fair 2025

  2. From Chaos to Choreography: Multi-Agent Orchestration Patterns That Actually Work — Sandipan Bhaumik

    Learn the distinction between decentralized event-driven choreography and centralized orchestration, with LangGraph illustrating the latter.

    Sandipan BhaumikAI Engineer Europe 2026

  3. The Production AI Playbook: Deploying Agents at Enterprise Scale

    Learn how measurable business outcomes such as query deflection guide an enterprise agent rollout, and where MLflow LLM judges and Unity Catalog fit.

    Sandipan BhaumikAI Engineer Europe 2026

Messages from the stage

Separate application design from model optimization

Khattab distinguishes scaling search and learning from engineering purpose-built systems. He advocates modular abstractions and interchangeable models so applications can benefit from new optimization methods without accumulating prompt-specific complexity.

Make agent failures recoverable

Bhaumik uses a stale PostgreSQL-cache incident to explain inconsistent agent decisions. His coordination guidance includes immutable versioned state, explicit handoffs, circuit breakers, and compensating transactions.

Establish evaluation before selecting models

Bhaumik's retail banking chatbot case study places evaluation infrastructure ahead of model selection. Deployment readiness also involves governed data, operational ownership, and monitoring accuracy, response time, and customer satisfaction after launch.

Affiliations reflect each recorded session, not necessarily current employment.

Company sources · checked 2026-08-28