Fireside Chat: Simon Eskildsen on Building Turbopuffer
- Source
- youtube.com
- Author
- AI Engineer
- Date

Context
Turbopuffer's early search engine combined object storage with a cache to address two different needs: durable storage and fast retrieval. In the fireside conversation, founder Simon Eskildsen describes grouping numerical document representations into clusters, storing those clusters as files, and fetching the groups nearest a search query.
That first implementation still faced the delay of fetching remote objects. He describes adding a cache in front of the object store so previously fetched files could be reused. The account explains how the prototype paired persistent backing storage with faster repeat access; it describes the early design, not the company's complete present architecture.
“I hate benchmarks so much because that's not a satisfying answer to me”
Simon Eskildsen
“the entire internet runs on small towns in Bavaria. I'm convinced.”
Simon Eskildsen
“if you're wondering about the name, it's like if everything goes south, we can always pivot into vapes”
Simon Eskildsen
“when I don't know how to play a game I just play with open cards”
Simon Eskildsen
“you could do a million vectors for a dollar”
Simon Eskildsen
- 0:00Introduction and Simon’s early history with computers
- 3:02The International Olympiad in Informatics and early competitive programming
- 4:13How Simon was recruited by Shopify while still in high school
- 8:46Engineering challenges and scaling infrastructure at Shopify
- 14:56Decision to leave Shopify and the creation of the "napkin math" project
- 20:40The origin and technical motivations behind Turbopuffer
- 24:46Design challenges of building a database on top of S3
- 28:41Cursor becoming the first major customer
- 35:36The meeting with Jensen Huang and Nvidia’s push for GPUs
- 39:01The competitive reality of cloud infrastructure and CPU scarcity
- 43:06Philosophical perspective on venture capital and funding
- 51:45Building a remote-first culture with the "campfire" concept
videoMost of Your Docs Readers Are Now AI Agents - Hahnbee Lee, Mintlify & Lincoln Murr, Coinbase
videoIf Your AI Scores 100%, Your Eval Is Wrong - Ramakrishnan Lokanathan, eslams.com
videoYour AI Agent Wants Your Data. Here's How to Lock It Down - Jeff Smith, Oracle
videoEveryone Has the Same AI Models. Here's What Sets Agents Apart - Richmond Alake, Oracle
Checking sign-in…
Loading comments…

