AI System
Autonomous Engineering Analytics
AI-powered report generator that turns Git, Jira, and AppsFlyer data into strategic development insights.
Git + Jira sources
~$0.04 per report
Overview
Automated pipeline that collects development data from GitLab (commits, merge requests), Jira (tasks, sprints, time tracking), and AppsFlyer (mobile installs), then processes everything through LLM-based enrichment to generate interactive HTML reports with strategic analysis.
Problems Solved
- Manual report compilation took hours and introduced bias
- No unified view across Git, Jira, and mobile analytics
- Difficulty assessing task complexity and business impact objectively
- Executive summaries required senior engineering time
Architecture
Data collection layer (GitLab API, Jira API, AppsFlyer API) feeds into a processing pipeline (task structuring, LLM enrichment, metrics calculation). Output layer generates responsive HTML reports via Jinja2 templates. SQLite database stores all collected data with accumulation model.
Key Features
- Multi-source data collection: commits, merge requests, tasks, sprints, mobile installs
- Rule-based task structuring: ticket extraction, platform detection, type classification
- AI enrichment via Claude Haiku: difficulty assessment, business impact scoring, epic attribution
- Strategic analysis with extended thinking for executive summaries
- HTML reports with Chart.js visualizations
- Velocity metrics: cycle time, throughput, slowest tasks
- Bug quality tracking: creation/resolution flow, severity distribution
- Contributor tracking with alias resolution
Results
- Full report with AI analysis costs ~$0.04
- Covers iOS, Android, Backend, Web Extension teams
- Accumulation-based data model enables flexible period queries
- Reports generated as standalone HTML files — no hosting needed
Stack