He keeps founding the company that solves a giant's problem - then watching the giant buy it. Now he is doing it to search itself.
Issac Roth. Stanford-trained in multimedia design, then twenty-five years spent making infrastructure that developers don't have to fight.
Right now, Issac Roth is co-founder and CEO of Orama, an open-source company that started as a tiny search library and is turning into something stranger: a database with a large language model wired into its core.
The flagship project, Orama, is a complete search engine and retrieval pipeline small enough to ship in a web page - under 2kb, zero dependencies, written in TypeScript, runnable in the browser, on a server, or out at the edge of a network. The second act, OramaCore, is the ambitious one. Roth describes it as a fourth-generation database that fuses full-text search, vector search, and a built-in LLM interface with reasoning and action planning, plus a runtime to write custom agents that live on top of your own data.
The bet is simple to state and hard to build. Most "AI search" today is a stack of glued-together parts: an embedding model here, a vector store there, an LLM bolted on at the end, guardrails taped over the seams. Roth wants those parts to be one thing. An answer engine that thinks about your data instead of merely indexing it.
It is a fitting problem for him. He has spent his whole career collapsing complicated infrastructure into products that feel like one clean idea. Search is just the latest layer of plumbing he has decided to pull up and redo.
Cisco to OpenShift to Node.js to LLM-native search. Read his resume and you are reading the history of enterprise computing - because he kept arriving a few years before everyone else.
Starts as a networking software engineer at Cisco, then becomes an early member of Wily Technology - the application-monitoring company whose alumni would go on to found New Relic and AppDynamics.
Co-founds and runs Makara, a cloud platform-as-a-service startup, back when "PaaS" barely existed as a word.
Red Hat acquires Makara and Roth stays to create and scale Red Hat OpenShift - giving himself the title "Red Hat PaaS Master" along the way.
Incubates core Node.js contributors at Shasta and becomes founding CEO of StrongLoop - betting on Node before it was the default.
StrongLoop is acquired by IBM; Roth becomes CTO for API Management and Hybrid Cloud Integration, shaping IBM's API platform.
Joins Shasta as Venture Partner and Managing Director, co-building the go-to-market accelerator Elevate and backing a roster of infrastructure startups.
Returns to building. Takes the CEO seat at Orama to reinvent search for the era of AI agents and copilots.
Presents "OramaCore: A Search Database with LLMs Built-In" at Data Council, putting the LLM-native database thesis in front of the engineering crowd.
One of the first serious platform-as-a-service offerings, grown out of his startup Makara. A foundation thousands of teams still deploy on.
Founding CEO of a company that became a commercial backbone of the Node.js ecosystem, then was acquired by IBM.
As an IBM CTO for API Management and Hybrid Cloud Integration, he helped define how a giant exposes its services.
At Shasta and as an angel: Scalyr, Apteligent, Beautiful.ai, CodeFresh, Outerbounds, Voxel51, CircleCI, Cloud9. A taste for picks-and-shovels.
Co-founder of a security and observability company - because the man cannot, it seems, stop starting things.
An open-source search engine that fits in a tweet's worth of bytes, and a database with an LLM at its center.
Here is the detail that explains more than any title. Roth's Stanford degree is not in computer science. It is in Multimedia Design & Technology.
That is an odd starting point for someone who would go on to wire together container platforms, API gateways, and vector databases. But it tracks. He describes himself as blending technical depth with design sensibility - and you can see it in the work. OpenShift, StrongLoop's tooling, Orama's two-kilobyte footprint: each is a piece of gnarly infrastructure dressed up to feel obvious.
The best infrastructure people are not the ones who love complexity. They are the ones who can't stand it - who keep folding the mess down until what's left looks like a single good idea. A designer's instinct, applied to plumbing.