Tushar Madan
Founder, Avianna · Product Adoption Lead, Databricks
I’ve spent a decade getting AI and data systems into production inside real enterprises. At Databricks I lead product adoption for Databricks Apps, working with teams putting agents to work. Avianna is where those lessons become frameworks, prototypes, and writing.
Advisory
When an AI strategy stalls between competing goals, I build clarity and alignment. A few startups and enterprise teams at a time. Email me.
Areas
Speaking
Talks on getting an organization aligned on one AI strategy it can actually execute. From the field, not the deck.
Formats
Recent
Data + AI Summit 2026, San Francisco (June 2026): spoke on “You Built That with Databricks Apps?! A Customer Showcase of What’s Possible” — what separates data apps and agents that reach production from prototypes that stall. Speaker spotlight ↗
Speaker bio
Tushar Madan is the founder of Avianna and a product adoption lead at Databricks. He has spent a decade deploying AI and data systems inside real enterprises, and speaks about what it actually takes to get them adopted.
He writes about AI systems, agents, and the mathematics underneath at avianna.ai.
To invite him to speak, email me.
What I work on
Enterprise AI adoption
How organizations get from AI demos to durable operating models.
Agent governance
Authorizing, tracing, and reversing agent work: the questions behind Concord and Lattice.
The field ↔ product loop
How field signal should shape what the platform builds next.
Background
A decade across the data and AI stack: Databricks since 2019, from solutions architecture through field-engineering leadership to product. Before that, ML and big-data architecture at Atos and FINRA. The formal version is on LinkedIn.
Writing elsewhere
On the Databricks blog: how the Minnesota Twins scaled pitch-scenario analysis to 20,000 simulations per pitch across 15 million pitches (part 1, part 2). On Medium: reproducing GPT-2 (124M) and a systems approach to navigating model drift.
In the press
Cool People Stories profiled the road from baseball simulations to agent governance (August 2026).
More at the essays · avianna.ai