Work

A cobbled Victorian warehouse street in London, with iron overhead walkways linking the brick buildings and a single figure walking between them.

The index of what I build, write, and publish. Long-form essays live here on the site; technical write-ups are on dev.to, where they're canonical, and are collected below.

Essays: AI and the Industrial Revolution

A series on the parallels between the Industrial Revolution and what agentic AI is doing to software engineering, argued from economic history and evidence, not prediction. New parts published as they're ready.

→ Read the series

  1. Cheap Tokens, Cheap Attempts. Agents don't out-think your engineers; they collapse the cost of an attempt. Why that makes judgment the scarce thing, what cheap coal and the spinning jenny have to do with it, and why the skeptical engineer is the asset adoption runs on.
  2. The Terrain Determines the Agent (forthcoming). You don't get a better agent by buying a better model. You get one by preparing the ground it runs on: harness engineering, test suites as evaluation that scales, and why clean benchmarks flatter a model that real codebases expose.
  3. The Factory Was the Real Invention (forthcoming). The machine sets the possibility; the organisation sets the distribution. What the labour-market evidence actually shows, why centralised approval becomes the bottleneck once attempts get cheap, and why the deciding variable is organisational design.

Open-source build: a coffee-roasting agent

An end-to-end project for detecting coffee first crack from audio and using it to help control a home roaster. It's the running example behind much of my writing on agentic ML: an open Hugging Face-native first-crack dataset and model, an ONNX INT8 export benchmarked on a Raspberry Pi 5, a live demo, and an MCP server that owns the roaster and session boundary. Current metrics, dataset counts, and benchmarks are maintained on the model and dataset cards.

Writing

Technical posts are published on dev.to/syamaner and remain canonical there.

Spec-driven ML: rebuilding the coffee agent (2026)

A six-part series on rebuilding the prototype into a production system, directing coding agents through a multi-phase ML project while keeping ownership of architecture, ML decisions, and quality gates. It ends with a live roast: four agents, one MCP server, and a supervised run on real hardware.

  1. The architecture and the agent
  2. Building the audio dataset
  3. The science: tuning to high precision
  4. Optimising an 86M-parameter audio transformer for Raspberry Pi
  5. From local model to live demo on Hugging Face
  6. Roast day: four agents and a live roast over MCP

The original prototype (2025)

How the project started: training the first-crack detector, building MCP servers for roaster control, and closing the loop with an agent.

  1. Training a neural network to detect first crack from audio
  2. Building MCP servers to control a home roaster
  3. Orchestrating MCP servers with .NET Aspire and n8n

Applied AI and RAG with .NET Aspire (2024–2025)

Earlier engineering writing (2022–2024)

Research

MSc Artificial Intelligence, University of Bath (distinction). Earlier research at the Pattern Recognition and Image Analysis group, University of Salford.

  • Data Mining Approach to Implement a Recommendation System for Electronic Tour Guides. EEE 2005.
  • Mining GPS Logs to Augment Location Models. WIT Transactions on Information and Communication Technologies, 2005.
  • Determining the Locations Visited by GPS Users: A Clustering Approach. CISST '04.

Profile and outputs: Semantic Scholar · Salford repository

Elsewhere