Temporal Data Provenance
Provenance has two axes: where a fact came from, and when it was true. AI needs both to reason reliably.
Member of Technical Staff · Doubleword
I'm Jaedon, a software engineer working on agentic AI, search, and data infrastructure. I'm currently Member of Technical Staff at Doubleword, where I drive enterprise integrations and partnerships for its high-throughput AI inference API, and the founder of Flux Search, a freshness-first semantic search engine for developers and AI agents.
Before this, I built the investor-startup matching engine and its data infrastructure at Evvolve & Partners, shipped AI tooling and ran OEM partnerships at ChipHub (NVIDIA Inception), sourced startups and ran due diligence as a venture capital analyst at Spartech Ventures, and led product at Certie, an Oxford University Innovation-backed startup.
A web search engine that organises the web by time, built for the era of agentic search. We call it 4D search: organise the past, and find what's next, now.
View project →
A simple website that alerts UK users when the Northern Lights are visible, using real-time aurora monitoring and email notifications.
View project →
A comprehensive financial analysis platform for stock traders with agentic AI workflows, real-time data visualisation, and portfolio management tools.
View project →
An interactive webapp where students pledge not to carry knives, visualised as a growing network graph of interconnected promises.
View project →Provenance has two axes: where a fact came from, and when it was true. AI needs both to reason reliably.
Genuine UK control over AI rests on open-weight models it can self-host, run, and inspect.
My first investor-startup matching engine produced confident, useless results. Here is what actually fixed it.