London --°
Shop
--:--:-- — 51.5074° N, 0.1278° W

Python Developer·Data Scientist·AI & ML

Shaz Moghaddam

Where data science, code, and product thinking meet, and turn into things people actually use.

Available for freelance, contract, and full-time opportunities, in London or internationally
50+
Projects
7
Apps
26
Digital Products
10+
Publications
6
Certifications

Projects

The commercial products below are built and published under Moghaddam Labs.

Python FastAPI PostgreSQL SQLAlchemy Alembic React TypeScript Vite Auth0 HMRC MTD API Playwright GitHub Actions Docker Render

A Making Tax Digital platform for UK landlords that keeps property records and files cumulative quarterly updates straight to HMRC. Landlords add income and expenses by hand or import a bank statement CSV through a two-step preview that blocks duplicate rows. Every figure is stored as integer pence and mapped to HMRC's property headings, with joint ownership shares and mortgage interest handled the way HMRC requires. It connects to HMRC over OAuth with encrypted tokens, sends validated fraud prevention headers, checks obligations and deadlines, and keeps an append-only audit trail of every submission. Built as a modular monolith (FastAPI, PostgreSQL, SQLAlchemy 2.0, Alembic) with a React and TypeScript frontend whose types are generated from the API, Auth0 sign-in in the UK region, and a CI gate that runs more than 600 automated tests, covering the backend on SQLite and Postgres plus frontend unit and browser tests with accessibility checks at phone, tablet and laptop sizes. The live demo runs end to end on HMRC's sandbox.

Live Demo
Python FastAPI NumPy rasterio Shapely Next.js TypeScript MapLibre Bezel Charts Geospatial Docker Google Cloud Run Vercel Playwright

A live web app that shows how exposed the area around a London home is to climate-related perils, before the insurance quote arrives. Search a postcode and mark the home on the map to get a risk band for flooding from rivers and sea and from surface water, the distance to the nearest zone in each band and the zones drawn on the map, with plain-English explainers for subsidence and summer heat. The Environment Agency's flood data is turned into 5 m grids and checked against London's boundary, so a gap in the data shows as "No data" and can never look like a low risk. Each region and peril is a plug-in that must say what its result means, how sure it is, what to do next and where its data came from, so adding the rest of England means adding data, not rewriting the core. It's private by design: no cookies or tracking, locations kept out of web addresses and logs, and postcode lookups routed through the server so the lookup service never sees who's searching. The FastAPI service runs on Google Cloud Run in London and the Next.js site on Vercel, built to WCAG 2.2 AA, with charts drawn by Bezel Charts and 178 Python and 139 website tests plus Playwright end-to-end tests.

Live Demo    View on GitHub
Python FastAPI Plotly Dash scikit-learn Causal AI SQLite Docker Claude AI Railway Render

An enterprise energy intelligence platform for UK and European industrial operations. It provides real-time multi-site energy monitoring, ML anomaly detection with an isolation forest model, causal AI root-cause analysis with quantified causal strength scores, and CSRD-ready ESG reporting covering Scope 2 emissions, GRI 302-1 and carbon intensity. Includes a plain-English AI assistant powered by Claude, JWT authentication with role-based access control, and 911 passing tests. The live demo is deployed on Render.

View on GitHub    Live Demo
Python FastAPI Claude API Pydantic SQLite Docker REST API OpenAPI Render

Structured extraction and deterministic validation for UK commercial energy bills, and the bill-intelligence module of VoltEdge. It accepts PDFs and phone photos (including iPhone HEIC), sends each one to Claude for schema-constrained extraction into a fixed Pydantic model, then runs a registry of independent validation checks covering MPAN digit counts, subtotal and VAT reconciliation, estimated reads, duplicate charges, contract-renewal windows and non-energy-bill detection. It ships a FastAPI review app with a drag-and-drop frontend, SQLite persistence, API-key auth and JSON, Excel and CSV export, plus a cross-portfolio layer that surfaces renewals, quantified suspected overcharges and supplier comparisons. Includes interactive OpenAPI docs, a Render deployment blueprint and 50 passing golden-file tests anchored on real, anonymised supplier bills.

View on GitHub
Python pandas Bezel Charts Eurostat FRED Our World in Data GitHub Actions pytest Open Source

A public data challenge on GitHub about Europe's energy since 2010: what it costs, where it comes from, who pays and how fast it's changing, through two shocks, Russia's invasion of Ukraine in 2022 and war in the Gulf in 2026. One script downloads the latest figures from Eurostat, FRED and Our World in Data, and a State of Europe's Energy report of eleven findings is generated straight from the data, so every number and sentence stays true as new figures arrive. At launch in November 2026 it showed Russia's share of the EU's imported gas falling from 41% to 11%, gas reaching 11.8 times its normal price in August 2022, almost no link between reliance on Russian gas and higher household bills, and wind now making more of the EU's electricity than gas. It also covers €651 billion of government support and the energy companies' record profits. A weekly GitHub Actions job tests the new data and opens a pull request with refreshed charts. All 25 charts are drawn with Bezel Charts, my own library, and each month brings a new brief with beginner to expert levels.

View on GitHub
Python pandas Bezel Charts ONS Open Data HMRC GitHub Actions pytest Open Source

A public data challenge on GitHub about UK jobs, unemployment and pay. One script downloads 33 ONS labour market series and HMRC payroll data, and a State of UK Jobs report of eight findings is generated straight from the data, so every number and sentence stays true as new figures arrive. At launch in October 2026 it showed long-term sickness up 620,000 since Covid to 2.74 million people, youth unemployment at 15.6%, a jobs market cooling from its 2022 record, and survey and payroll figures pointing in opposite directions since late 2024. A weekly GitHub Actions job tests the new data and opens a pull request with refreshed charts after every ONS release. All 15 charts are drawn with Bezel Charts, my own library, and each month brings a new brief with beginner to expert levels.

View on GitHub
Python FastAPI Plotly Dash SQLAlchemy scikit-learn spaCy SQLite Docker Claude AI Render

A B2B clinical trial site performance intelligence platform for mid-market CROs in the UK and EU. It monitors enrolment velocity, risk scores and protocol deviations across trial sites, with an AI-assisted insights layer powered by Claude. Features a FastAPI backend, a Plotly Dash dashboard and a SQLAlchemy data layer that seeds itself on startup. Built across five phases with 496 passing tests.

View on GitHub
Python Flask React pandas NumPy SciPy Docker AWS / GCP ONS Open Data

A production-grade UK salary intelligence platform built on ONS open data. It aggregates Annual Survey of Hours and Earnings data, applies CPI inflation adjustment and statistical analysis, and delivers fair salary benchmarks through a REST API and a React dashboard. Features include regional breakdowns, gender pay gap trends, occupation distribution profiles and CAGR analysis, all GDPR-compliant with no personal data collected. Deployable to AWS ECS Fargate or GCP Cloud Run with a multi-stage Docker build.

View on GitHub

What I've Shipped

Web Components SVG JavaScript Python npm PyPI Data Visualisation Open Source

Sixteen zero-dependency chart tiles drawn in pure SVG, a sibling of Bezel: sparklines, lines with real date axes and log scales, stacked areas, bars, bars and a line on two axes, waterfalls, histograms, box plots, a scatter plot with a least-squares trend line and Pearson's r, rings, donuts, gauges, progress bars, bullet charts and a year-long heatmap. Each chart is one HTML tag that fills its container, follows the viewer's light or dark setting, formats numbers for any locale, works with the keyboard and describes itself to screen readers. Data can be set straight from JavaScript, and clicks on any bar, point or slice fire events for React and Vue. A Studio imports CSV, Excel or JSON, or cells pasted from a spreadsheet, and exports the chart as HTML, a shareable link or a static SVG for READMEs, Notion and slides. It's published on npm (npm i bezel-charts) and as a Python package on PyPI (pip install bezel-charts) that draws the same charts from pandas DataFrames, tested byte for byte against the JavaScript across 257 chart cases in CI. Both always ship the same version, released from a Git tag. About 26 kB gzipped for all sixteen, or 7 to 12 kB for a page that loads just one.

View on GitHub    Live Demo    View on npm    View on PyPI
JavaScript HTML & CSS PWA Service Worker Python pytest GitHub Actions Open Source

A free web quiz with 1,200+ questions in 18 subjects, from capitals and cars to cartoons and Champions League finals. Every answer comes with a short fact and its source: country and science facts are only used where two independent datasets agree, maths is worked out by code, and everything else was checked against a published source, with an as-of date wherever a fact can change. Questions are higher or lower, yes or no, or pick one, with four ways to play: a daily ten that's the same for everyone, quick rounds, three lives, and Hunch-off for up to six players on one phone, with 100 original jokes between rounds. It works offline as an installable web app and updates itself, with dark and light looks, sound, personal stats, full keyboard play and screen reader support, and no accounts, adverts or tracking. Plain HTML, CSS and JavaScript, with the question bank built in Python and checked by 51 tests on every pull request.

View on GitHub    Live Demo
Web Components SVG JavaScript Container Queries Open Source

Five zero-dependency web components for clocks, a calendar, live weather, an at-a-glance dashboard and a large-type poster calendar, inspired by Apple's StandBy widgets. Every tile is drawn in vectors and sized with container query units, so it looks the same at 120 px or across a 4K screen. A Studio lets anyone design a tile, share it as a link and copy the code, with automatic dark mode, live weather from Open-Meteo, .ics calendar feeds and weather words in seven languages. About 25 kB gzipped, tested with unit and browser suites, with zero axe accessibility violations.

View on GitHub    Live Demo
HTML5 Canvas Vanilla JS Generative Design Open Source npm

An open-source JavaScript library of twenty morphing dot animations, from a Fibonacci sphere to a vortex, an iris and a figure of eight. Any shape morphs smoothly into any other, drawn as dots or as glowing lines, and the orbs can move to music or a microphone. Named states such as thinking, success and error make it simple to build loaders and AI assistant indicators. Design an orb in the Studio and export it as MP4, GIF or PNG, or run it full screen as a desk display. Available on npm (npm i @shazmoghaddam/orbs) with TypeScript types and React and Vue examples. One file, about 11 KB gzipped, no dependencies, with support for reduced motion, keyboard use, and light and dark themes. Try it in the Get in Touch section below: click the orb to change its shape.

View on GitHub    Live Demo    View on npm
Python fontTools Shapely SVG Vanilla JS Generative Design Open Source

An open-source kit of 183 layered-line symbols, letters and numbers, inspired by the stacked layers of a 3D print. A Python pipeline built on fontTools and Shapely extracts glyph outlines from Inter Display, resolves nested contours and generates original geometric shapes. A custom JavaScript scanline engine then slices every outline live in the browser at any angle, merges overlapping letters and tests several grid offsets to remove stray slivers. Designers can tune line count, weight, angle and four effects, colour symbols with a solid or gradient ink, set any word in the type tester, and export SVG, PNG, CSS or a ZIP with an SVG sprite.

View on GitHub    Live Demo

What I've Built

Android Flutter Google Play Store Free

A higher-or-lower guessing game inspired by higherlowergame.com. Players see two items and guess which one gets more monthly Google searches. It has 3,187 questions across 11 categories, including Sports, Celebrity, Tech, Gaming, Geography and History, with 3 lives, high scores per category, subcategory navigation and smooth card animations. Free to play.

View on Google Play
Android Google Play Store Free

Build a world from nothing and watch it grow. World Builders is an open sandbox that gives you the tools to shape the land and expand it piece by piece, at your own pace. No timers and no fixed goals, just room to create and experiment. Free to download and play.

View on Google Play
Android Google Play Store Ages 3–11 Free

An adventure-based learning app for children aged 3 to 11. It combines storytelling and exploration with maths challenges to build core skills in a way that feels like play rather than study. Free to download and use.

View on Google Play
Android Google Play Store Free

A fun, bingo-style maths game for Android. Familiar bingo mechanics make practising maths engaging, and the design keeps players coming back. Free to download and play.

View on Google Play
Android Google Play Store Free

A minimal, fullscreen animated live wallpaper for Android that shows the current month. Fully customisable, with multiple languages and fonts, a breathing animation and a live clock. Free to download.

View on Google Play
Android Google Play Store Free

A minimal, fullscreen animated live wallpaper for Android that shows the current day of the week. Supports multiple languages and fonts, with a breathing animation and a live clock. Free to download.

View on Google Play
Android Google Play Store Free

Turns an Android phone into a living departure board. Inspired by the classic split-flap Solari boards, it shows live world clocks for multiple cities with authentic flip animations. Supports 300+ world cities and multiple colour themes. Free to download.

View on Google Play

Skills & Tools

  • Python
  • Kotlin & Java
  • Flutter
  • JavaScript
  • TypeScript
  • HTML & CSS
  • SVG
  • SQL & PostgreSQL
  • SQLite
  • Machine Learning
  • NLP & spaCy
  • pandas & NumPy
  • scikit-learn
  • TensorFlow
  • SciPy
  • Data Visualisation
  • Geospatial (rasterio, MapLibre)
  • FastAPI
  • Plotly Dash
  • React
  • Next.js
  • Flask
  • SQLAlchemy & Alembic
  • Pydantic
  • Vite
  • Node.js
  • Web Components
  • Electron
  • Streamlit
  • Docker
  • Git & GitHub
  • GitHub Actions (CI/CD)
  • Publishing to npm & PyPI
  • pytest, Vitest & Playwright
  • REST APIs
  • Jupyter
  • Matplotlib & Seaborn
  • Plotly
  • Tableau
  • fontTools & Shapely
  • Vanilla JS
  • HTML5 Canvas
  • AWS
  • GCP & Cloud Run
  • Vercel
  • Railway
  • Render
  • Auth0, OAuth 2.0 & OIDC
  • HMRC Making Tax Digital APIs
  • Accessibility (WCAG, axe)
  • Causal AI & NetworkX
  • Android Development
  • Generative Design
  • UI / Visual Design
  • Digital Product Design
  • Open-Source Maintenance
  • Anthropic Claude API
  • ChatGPT & Gemini
  • Photography
  • Videography

From raw data
to something useful

I'm a London-based Data Scientist and Python Developer, trained through Imperial College London's Data Science Bootcamp (96% average). My work covers the whole data workflow, from cleaning messy datasets and building models to turning the result into a tested, deployed product. Lately that has mostly meant B2B platforms, like a Making Tax Digital service that files to HMRC, a climate risk checker for London homes and an energy intelligence tool for industrial sites. I publish them under Moghaddam Labs, which I founded in 2022.

Outside client and product work, I build my own things: Android apps, generative art systems, live wallpapers, word-cloud typography. Fifty-plus projects so far, spread across the Google Play Store, Gumroad, Envato, and Lemon Squeezy. Running them myself taught me things that are hard to learn any other way: how to price something, when to ship versus keep polishing, and what users respond to once a product is out in the world.

There's a quieter part of my life too. After major surgery and the birth of my eldest child, I started donating blood regularly, and have now made 50+ donations, with my 50th recognised by NHS Blood and Transplant. It's a small, repeatable way of putting something back.

I also write, though not often and not because I'd call myself a writer. On Reddit, I share programming work and what I've learned building it. On Medium, I write personal essays, because some thoughts just need somewhere to land. You'll find those in the Writing section below.

Shaz Moghaddam Shaz Moghaddam
What I Do

Certifications

Drag to explore

What I've Written

Reflections on becoming, connection, and the things that stay with you.

Get in Touch

Open to freelance, contract, and full-time opportunities, in London or internationally, wherever the work is interesting. If you're building something that matters, bring the problem. I'll bring the code, the models, and the thinking.

Download CV
Available on
Google Play Gumroad Envato Lemon Squeezy Medium DEV GitHub GitHub Pages jsDelivr npm PyPI Instagram Reddit LinkedIn Render Vercel
Built with
Python pandas Matplotlib NumPy Jupyter scikit-learn spaCy Plotly Bezel Charts Tableau FastAPI Flask React Next.js JavaScript TypeScript Web Components SVG Node.js Vite PostgreSQL SQLite SQLAlchemy Alembic Pydantic pytest Auth0 GitHub Actions Playwright Flutter Docker Google Cloud fontTools Shapely rasterio MapLibre Claude ONS GOV.UK Data.gov Environment Agency
Made with intention