Lucas Milanez

Multi-tenant SaaS that was published on the monday.com marketplace (now archived). Solved a real platform limitation: monday.com cannot trigger automations at exact times. It monitored DateTime columns and fired actions (status change, notification, webhook) at the precise moment configured by the user, with millisecond-precision scheduling.

• Multi-tenant architecture with full data isolation by account_id
• Separate worker process from web server (independently scalable)
• OAuth 2.0 tokens encrypted at rest (AES-256-GCM), HMAC webhook verification
• BullMQ job scheduling with Dead Letter Queue and graceful shutdown with queue draining
• Per-tenant rate limiting (Redis sliding window), orphaned trigger detection, data retention policies
• Property-based testing (fast-check) on critical business rules, Vitest + Testing Library
• Integrated billing via Stripe with free trial and per-account limits
• Infra: Docker multi-stage build, Caddy reverse proxy, Hetzner VPS, automated daily backups

Open-source MCP server that turns Microsoft 365 into context for any AI client (Claude Desktop, Cursor, Kiro). Replaces Microsoft Copilot ($30/month) at ~$1/month with semantic cross-service search, persistent memory, and automatic pre-meeting briefings generated by Claude Haiku 4.5. Exposes 9 MCP tools over JSON-RPC 2.0 (spec 2025-06-18).

• 9 MCP tools with JSON-RPC 2.0 handshake (initialize, tools/list, tools/call)
• Semantic search across Calendar, Mail, OneNote, OneDrive in <2s (all-MiniLM-L6-v2, 384d)
• Pre-meeting briefings via Graph webhooks + Claude Haiku with prompt caching
• Persistent memory (save_memory / recall_memory) across sessions and clients
• Append-only audit trail, OAuth 2.0 + PKCE, Fernet AES-256 token encryption
• 100% open source · Fly.io + Vercel + Neon deploy at ~$1/month

Suite of 3+ internal web applications that replaced manual spreadsheet workflows for 20+ non-technical users in production and operations. Reduced recurring reporting tasks from hours to minutes with browser-based tools for request tracking, incident reporting, and daily operations management.

• 3+ apps used daily by non-technical teams
• REST APIs with Python backend logic and SQL integration
• Relational database modeling for business workflows
• Reduced manual reporting effort across departments

Executive dashboard consolidating production metrics and commercial data from 4+ sources into a single view. Built with Power BI connected to Python ETL pipelines, cutting weekly report preparation from ~3 hours of manual work to automated, real-time delivery.

• Consolidated production and commercial KPIs
• Automated ETL pipelines replacing manual data prep
• Interactive drilldowns for leadership decision-making
• Reusable Python scripts for recurring reports

Interactive analytics dashboard processing 10k+ rows of coffee sales data with automated ETL, revenue forecasting, and visual storytelling. Helps stakeholders compare market performance and spot growth opportunities in seconds.

• Real-time KPIs and drilldown exploration
• Comparative analysis across Melbourne & Sydney markets
• Dark-themed UI with interactive heatmaps and trend charts
• Exportable insights built with Streamlit