Julian Cano

🔥 Challenge

Tracking job vacancies and judicial auctions published across dozens of government sites and private portals with no APIs, each with different formats and publishing rhythms. Thousands of records per week appeared duplicated across sources and needed deduplication, classification and daily delivery via WhatsApp to a large recipient list without getting blocked or requiring manual curation.

🔥 Challenge

Depending on SEO SaaS meant recurring costs, quota limits and zero control over the data. The goal was a self-hosted toolkit exposing its tools via MCP so AI assistants could query them directly, with own monitoring and no third-party stack.

🔥 Challenge

Entering the cloud industry with a verifiable standard instead of pure self-teaching: taking official training and applying it immediately to real project infrastructure, not throwaway labs.

🔥 Challenge

Centralizing mass driver recruitment by migrating fragmented WhatsApp flows into a unified web channel. It demanded real-time identity validation capped at 3 attempts, zero re-entry of data for re-applicants, detection of work history and rehiring across 3 legal entities simultaneously, and strict legal traceability under Chilean Law 21.719.

🔥 Challenge

Replacing real animal dissection in schools with an immersive VR experience that runs on affordable educational hardware. The technical challenge: realistic hand and tool interactions that felt surgically precise, and anatomically faithful specimens that let students learn without harming living beings.

🔥 Challenge

Giving artificial intelligence to NPCs in Unity VR environments: characters that speak with natural voice, understand the interactive world, generate dynamic images and coordinate multiplayer sessions — all from one unified backend instead of per-scene scripts.

🔥 Challenge

A B2B SaaS business intelligence platform with two critical problems: a 4500-line frontend monolith running a silent loop firing thousands of unnecessary requests, and a backend with inefficient sequential SQL (12 queries to compute a single average). On top of that, AI-driven dynamic KPI creation (Text-to-SQL) posed risks of hallucinations, destructive SQL injections and multi-tenant data exposure.

🔥 Challenge

I needed a portfolio that reflected my real stack — AI, automation, full-stack — and that any dev could clone and have running in minutes. Most templates are generic, ship dead placeholder data, or simply lack a CMS: changing one text means editing code and redeploying.