Vol.  04 No.  01
Issued  April 2026
Filed from  London
jcousins/2026-04-29

A working journal of applied engineering

Jevan Cousins

Applied engineer.


Author

J. Cousins
Independent · London, UK

Keywords

applied AI · data engineering · forward-deployed systems · cross-domain product · independent research · adaptive learning

Abstract. This is the personal journal of an applied engineer. The author was trained as a physicist (UCL); spent four years architecting AI systems and data pipelines inside a €176B European asset manager (Allianz Global Investors); and is now building independently across live music, nutrition, career systems, personal AI, equity research, and SMB AI discovery. The thesis running through the work is that real-world impact comes from going deep on a domain, understanding the problem on its own terms, and shipping the technology that encodes the understanding. Each paper in this issue documents an active artefact in that programme.

Aims To document the work openly, make the decisions and trade-offs visible, and invite correspondence with anyone working on adjacent problems.

In figures

€176B

Fixed Income AUM platform supported at AGI

7

Working papers in this issue

1M+

Live-music events ingested daily for Lineup

12 mo.

Production uptime, zero data loss

Folio · II

From the reading desk

Currently reading

The Scaling Era

Dwarkesh Patel · 2026

A first-hand chronicle of the people, decisions, and capital flowing into frontier AI. Reading slowly because the bibliography is half the value.

Active curriculum

Agentic AI AI Use Cases CS Fundamentals Data Engineering Production ML & MLOps Quantitative Finance System Design
To the reading room →

Folio · IV

In preparation

Three literature reviews on the desk: a working review of wealth inequality and the evidence on social mobility; production ML observability; agentic AI evaluation in 2026. Plus three editorials in draft, including a longer note on leaving a job I was good at.

Notes & literature reviews →

Folio · V

Archive

Earlier work and the bedrock chapters. The UCL Physics dissertation on deep learning for image reconstruction in lithium-ion batteries; the DataCamp Data Scientist certification; long-running mentoring; sport for Sussex; Young Enterprise. Bedrock for everything since.

Past work archive →

End of issue

If a problem you can't quite get out of your head sits at the boundary between a domain and a technology, write to me.

For collaborations, conversations, or quiet introductions. Roles too, where the description and the work above genuinely overlap. I read everything and reply to most.