Shaz Moghaddam

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.

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.

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.

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.

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.

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.

A public monthly data challenge on GitHub that asks how UK house prices respond to economic shocks. It's built on the HM Land Registry UK House Price Index, with monthly prices from 1995 to 2026 by country, region, local area and property type. The first brief tracks the 2008 crash region by region: London took about 4 years to get back to its peak, the North East 13.5, and Northern Ireland is still below where it was in 2007. It also covers the stamp duty holiday spike and the 4% dip as Bank Rate rose from 0.1% to 5.25%. It comes with six reproducible charts, a pandas starter script and a roadmap for joining prices to ONS affordability, Bank Rate and CPI data. Anyone can take part, with a GitHub Discussion for sharing findings and voting on topics.