Legal Document Management Platform

A comprehensive legal tech platform enabling law firms to automate document processing, client onboarding, and compliance workflows with advanced AI capabilities.

Key achievements include implementing OpenAI powered semantic keyword validation with confidence scoring for legal document analysis, AI-driven client information extraction from retainer agreements, and OCR processing with word-level bounding box detection for accurate document highlighting. Built a dynamic questionnaire system with audio explanations to educate clients on legal terms.

Developed a multi-role platform serving clients, solicitors, and administrators with role-based access control. Created automated scheduling systems for usage monitoring, threshold alerts, and monthly billing report generation. Implemented real-time document funnel analytics, PDF manipulation pipelines (merge, convert, highlight), and an enterprise email notification system with retry logic and delivery tracking.

Architecturally it is a TypeScript Express API in front of Firestore and Firebase Storage, with Firebase Identity Platform handling authentication and SMS multi-factor for solicitor accounts, and a React and Material UI front end that renders documents in the browser, captures electronic signatures on canvas and drives the guided client journey. The document pipeline is the awkward part of the build. Firms send whatever they have, so incoming files are normalised before anything else can happen: office documents are converted with a headless LibreOffice installed into the deployment image, scanned pages fall through to OCR, and extracted text is matched back to word level bounding boxes so a clause can be highlighted in place rather than merely quoted. Pre-highlighting pages ahead of the client opening them keeps the viewer responsive on long agreements.

Around that sit the operational pieces a multi tenant deployment needs: per tenant departments and role hierarchies, usage tracking with threshold alerts and scheduled reporting, an email log that records delivery outcomes and retries failures, session tracking with inactivity timeout, and bilingual content delivery. My own contributions concentrated on the document conversion and extraction path, the deployment configuration that makes it reproducible, and the Firebase data access and authentication layers.