Jevan Cousins

I am an applied engineer based in London. The label is plain on purpose: I prefer it to the more fashionable adjacent ones because it captures what I actually try to do, which is take a real-world problem in a specific domain (finance, science, business, my own life) and build the technology that turns the understanding of the problem into something useful.

My route to that work runs through physics. I did Natural Sciences at UCL, finishing with a first-class dissertation on deep learning for image reconstruction in lithium-ion battery Bragg coherent diffractive imaging. I cared more about the dissertation than the degree, which is probably the truest sentence in this paragraph.

I spent the four years after university as a Solution Architect inside Allianz Global Investors, reporting into the Global COO. The work was unusually broad for the title: architecting the firm's first enterprise agentic AI system; delivering MSCI BarraOne performance attribution at the third attempt after two prior failures over seven years, across a €176B Fixed Income platform; running a structured discovery that identified 50+ candidate AI use cases and shipped several to production; directing AI upskilling for 200+ staff. I left in February 2026.

---

AWhat I am working on now

Independently, in parallel: Lineup, a production data platform behind a SwiftUI live-music app; NutriPlan, a personalised meal-planning system on a retrieval-then-optimise architecture; JobHunter, a single-user system that runs my own job search; the earnings-call analyser, a research replication of recent work on Q&A alignment as an equity signal; the personal AI layer stitching OpenClaw and a Cowork-backed assistant; the adaptive learning system driving my own technical study; and an SMB AI discovery practice bringing the enterprise discovery method to smaller operators.

The thread, if there is one, is that I get pulled toward problems where the leverage comes from sitting at the boundary between a domain and a technology and refusing to leave until the work joins up. Real impact, in my experience, comes from going deep on the domain, not from running off-the-shelf tooling at it.

---

BSome things that are also true

I read widely and study deliberately. The Reading page lists what I am studying, what I am currently reading, and the broader subjects I keep returning to. I am particularly interested in wealth inequality and what the evidence says about social mobility; this comes from somewhere honest, not from a position paper.

I spent years tutoring and mentoring children from under-privileged backgrounds, mostly in maths and the sciences. Most of what I think I know about how people learn, I learned in those rooms. The adaptive learning system I built for my own study borrows heavily from that experience.

I played football, cricket, and long-distance running for Sussex at secondary school. Sport has been a constant: discipline, training cycles, and the useful habit of being wrong in front of other people. I still run.

I am a native English speaker and proficient in French (C1). I read in both, and I keep Lectures francophones on the Reading page on purpose, because the language fades quickly without use.

I went to the Databricks Data + AI World Tour in London this past cycle. Beyond that, my conference attendance is light by choice; I get more out of building and reading than out of programmes.

---

CAbout contact

This site is a working journal more than a CV; if anything here is the reason you are reading, please write. I read every message and reply to most. The right reasons to make contact, in roughly descending order of probability of a useful conversation: a problem in a domain you know well that you suspect benefits from the kind of work documented here; a paper or idea you'd like a second opinion on; a role where the job description and the work above appear to overlap. Bien à vous.