Software Engineer
February 2024 – June 2026
Most of my time here went on the three internal platforms below, and I was the main engineer on each. That ran from the Prisma data models behind the claims and document-collection apps through to auth and UI. Requirements were rarely written down, so a lot of the job was sitting with the staff who did the work and building from what they showed me.
R&D Claims Platform — Claim tracking and client management for the firm's R&D tax practice. Claims move through a seven-stage pipeline from prospect to HMRC payment, and the homepage shows where the money is: submitted, in time, overdue, queried or paid, with each figure opening the list of claims behind it. Client records are filled in from Companies House, which also flags clients with overdue filings. Staff can read and answer each client's HMRC email inside the app.
Document Collection Portal — A portal we built for collecting documents from the directors of newly acquired companies. Each director gets a private link and works through six stages covering ID, company accounts, HMRC details, creditors, debtors and assets. Staff approve or reject every field and file as it arrives, and uploads are filed straight into the right company folder on Nextcloud.
TARA, AI Narrative Assistant — A drafting tool for R&D claim narratives. An adviser answers a set of questions in rough notes and the app writes each section of the HMRC report from them, quoting the statutory definition of R&D where it matters and feeding earlier sections into later ones so the report stays consistent. Any section can be redrafted on its own, and the adviser picks which version ends up in the final Word document. The question set is configurable rather than fixed, so the same tool can be set up for other kinds of reports, legal work included.
Next.js React TypeScript Prisma MUI Recharts Node.js OpenAI REST APIs
