elliot.hegraeus
en sv

Skills

How I actually use these.

Not a checklist — where each tool earns its place in my work. Scroll, drag, or pick one above.

Strongest Language

01 / 14

Go

My default for backend services and CLIs — this site's server is Go. I reach for it when I want fast, boring, reliable concurrency.

As seen in
Language

02 / 14

Rust

For the performance-critical core: Quartz's native app and scheduling engine, and the compute behind my quant models. Where correctness and speed both matter.

As seen in
Strongest Language

03 / 14

Python

Quant work, data pipelines, and automation. FastAPI services, Monte Carlo models, and the glue that ties research to shipped tools.

As seen in
Language

04 / 14

TypeScript

Frontends and full-stack Next.js apps — typed end-to-end so the interface and the API agree.

As seen in

05 / 14

FastAPI

My go-to for Python APIs — used at Fairlo for reporting services and across my own backends when I want typed endpoints quickly.

As seen in
Strongest

06 / 14

PostgreSQL

The database I reach for first. Schema design, query tuning, and the system of record behind most of my products.

As seen in

07 / 14

Redis

Caching, queues, and low-latency state — Lootit's micro-betting core leans on it for fast reads.

As seen in

08 / 14

Docker

How I package and ship everything — consistent local dev and deploys across my services.

As seen in
Strongest

09 / 14

Backend systems

The throughline of my work: APIs, schedulers, sync engines, billing, and the plumbing users never see but always feel.

As seen in
Strongest

10 / 14

Automation

Making systems run themselves. At Abeloo I built a tool that cut a research workflow from an hour to a minute.

As seen in

11 / 14

Monte Carlo

Stochastic simulation for risk and pricing — built into real tools, not just coursework. See the Document Model and Complex Model projects.

As seen in

12 / 14

Regression & optimisation

Fitting models and tuning systems — regression-based reporting at Fairlo, optimisation in my quant projects.

As seen in

13 / 14

Machine learning

Applied ML where it earns its keep — learning metrics in Rebook, and ML-adjacent modelling rather than research for its own sake.

As seen in

14 / 14

Agentic workflows

Using AI agents to ship faster — but on top of a stack I learned the hard way first, so I know when to trust them and when not to.

As seen in