About

David Coldeira

Software engineer with 10+ years building production systems, and a physics degree behind the way I build them. Much of my work now involves LLMs given real tools and real data, then constrained so they can be trusted in production.

At work

I’m the lead software engineer on GQMLab, the laboratory information management system behind a UKAS-accredited geotechnical laboratory: a Python/Flask REST API, MySQL, a Flutter front end, Docker and Linux.

It includes Gio, an AI assistant with tool-use over the live system: SQL, charts and report-schema introspection behind a read-only database user, validated queries and role-gated access. An accredited lab is an unusually unforgiving place to put an LLM: every answer has to be reproducible and every query auditable.

Open-source work (own time)

Quantum Relational Language (QRL)

A quantum programming language where entanglement is a first-class citizen: relations-first, compiling to measurement-based quantum computing. Type system where Switch(d) programs are provably causally nonseparable; 968 passing tests; ~15.1k lines across 31 modules; four backends (Perceval, PennyLane, graphix, Qiskit).

Hardware-validated on a Quandela photonic QPU: S = 2.61 ± 0.08.

Most of the writing on this site is QRL research.

Bell

An assistant for QRL: plain English in, real QRL code executed in a sandbox, real numbers out. Hosted API or a self-hosted Qwen3-8B + LoRA fine-tune (815 instruction pairs). bell.entangledcode.dev

Froe

A local-first coding agent in Go: a single static binary with an agent loop, file and shell tools, adapters for local model runtimes (llama.cpp, Ollama, LM Studio, vLLM) plus Mistral (EU-hosted) as a cloud fallback, and a Neovim client. I use it to build QRL and Bell. GitHub

Background

Education: BSc Physics (Combined Sciences), Heriot-Watt University

Languages: Python, Go, Dart, SQL, JavaScript · English, Spanish, Italian

Stack: Flask, FastAPI, Django, Flutter, PyTorch, Docker, Podman, Linux, PostgreSQL, MySQL

Contact