I build automation tools that kill busywork — from Excel macros to full desktop platforms.
finsight Python · CustomTkinter · pandas · scikit-learn · SQLAlchemy — Enterprise lending analytics & automation desktop suite (NBFC/FinTech). 10 modules: executive KPIs with a Business Health Score, offline natural-language querying, SQL studio, reconciliation engine, one-click MIS packs, forecasting/risk analytics, and an audited task scheduler. 62 tests, CI on Ubuntu and Windows, ships its own synthetic lending book so it runs in 5 minutes.
visionqc Python · OpenCV · scikit-learn · FastAPI — Automated visual quality-control platform: classical CV defect detection (~2,500 img/s multi-threaded) + explainable ML classification (0.99 cross-val), REST API with audit trail, Docker, fully reproducible CI.
filesmith Python — Rule-based file organizer CLI: sort, dedupe (SHA-256), archive by YAML rules — dry-run by default, full undo journal.
xlforge VBA + xlforge-web TypeScript — Excel automation toolkit twice: desktop VBA (100k+ row cleaning, instant reports) and its Office Scripts + LAMBDA rebuild for Excel on the web.
Python · pandas · scikit-learn · OpenCV · SQLAlchemy · FastAPI · CustomTkinter · TypeScript · VBA · SQL · GitHub Actions · pytest
Dry-run by default — destructive actions are opt-in, never a surprise.
Every tool leaves an audit trail — automation you can't audit is automation you can't trust.
Measured, not assumed — benchmarks and tests over folklore.
If a non-developer can't tell what it did, it isn't done.
GitHub: @kristic8998