My contribution
- Browser-session and report extraction workflow
- Financial-year, date, pagination and checkpoint handling
- Canonical data model, matching and difference classification
- Readiness scoring, progress events and recovery tools
All projects
ScrapeForge
ERP-migration automation: reliable multi-year data extraction from legacy systems, then reconciliation you can trust.
My roleDesigned and implemented the system end-to-end, including the AI logic, backend, frontend, data processing and automation workflows.
Problem
Problem
When housing societies move to NoBrokerHood, years of financial records sit in legacy ERPs. Extracting them by hand is slow, and without systematic reconciliation nobody can be sure the migrated books agree.
System pipeline
- Source ERP
- Automation
- Extraction
- Normalise
- Reconcile
- Readiness
Approach
Validation is part of the system, not an afterthought.
Automates extraction of multi-year financial reports from legacy ERPs with checkpoints and count validation, then reconciles them against the new system. The workflow keeps inputs, decisions and exceptions visible so that a reviewer can understand how the output was reached.
Try the demo
ScrapeForge: interactive demo
Simulated legacy ERP → real filtering, pagination, extraction, normalisation, de-duplication and reconciliation.
- Computed
- Simulated visual
- Local model
Session ready
Create a synthetic browser session.
Reports Receipts Reconciliation
Receipts report2026-01-01 → 2026-09-30
Ready for a synthetic extraction run.
◇
Choose a date range to run real filtering, de-duplication and reconciliation.
This interactive demonstration is a simplified, synthetic representation inspired by an enterprise project. Production systems, company data and proprietary implementation details are not publicly exposed.
Result
A result that can be inspected.
The portfolio demo filters and paginates synthetic records, removes exact duplicates, reconciles two datasets and derives its readiness grade from the computed match rate.
Technical architecture
Stage-by-stage system design
- 01Source ERP · One browser, one session
Legacy ERP session → Adopted browser session
- 02Automation · Browser automation
Report + date range → Loaded, filtered report
- 03Extraction · Count-validated extraction
Paginated table → Raw dataset + checkpoints
- 04Normalise · Canonical financial model
Raw datasets → Canonical records
- 05Reconcile · Explainable reconciliation
Canonical records (old vs new) → Matches + difference categories
- 06Readiness · Migration readiness scoring
Reconciliation results → Readiness report
Tech stack
Tools used in the production system
- Playwright · Chromium
- FastAPI
- Pandas · RapidFuzz
- WebSockets · noVNC
The portfolio demo itself is static TypeScript running locally in the browser with synthetic data and no API key.