Internal QA tooling

Multi-Model AI & Risk-Scored QA Automation Engine

QA Tools — Create Test Card view, generating a Trello card with a grounded test plan from a pasted feature description

QA Tools — Release Summary view, reading the Trello board and generating release summaries plus a production changes log

The resulting Trello card — a generated test plan checklist, fully completed, tied to a real production feature

QA Tools — Create Test Card view, generating a Trello card with a grounded test plan from a pasted feature description

Overview

A risk-prioritization engine and serverless automation suite that cuts pre-test planning overhead by orchestrating multiple LLMs with resilient fallbacks.

What it does

Architected a 7-factor usage-calibrated risk-scoring algorithm with dependency-graph traversal across 41 SaaS modules, backed by a 122-case node:test suite that is estimated to cut manual test-planning prep from 30–45 minutes to under 5 minutes per feature.
Integrated Gemini, Claude, and Groq (Llama 3.3 70B) with exponential backoff, token-budget tuning, and prompt-calibration loops for multi-LLM resilience.
Developed a Telegram-to-Trello automation webhook on Vercel handling custom Unicode/bidi text parsing.

Tech stack

NNooddee..jjssVVeerrcceellPPllaayywwrriigghhttGGrrooqqCCllaauuddeeGGeemmiinnii

This is a private/internal project — no public link or repository is available.