Software Engineering student at the University of Seville, graduating in 2027.

🇬🇧 British citizen · Full UK right to work · No sponsorship required

I enjoy building software where backend engineering, AI and real-world product problems meet — from validating LLM-generated data before it reaches users to deploying multi-service applications in the cloud.

My main interests are backend systems, applied AI, integrations and cloud engineering.

An intelligent platform for professional tour guides combining AI-assisted route creation, geospatial validation, optimisation and live tour tools.

Django · Python · PostgreSQL · PostGIS · pgvector · LangGraph · Gemini · OR-Tools · Redis · Docker · Google Cloud

I worked as Product Owner in a 13-person engineering team while also contributing directly to the codebase.

Some of my main contributions included:

• Building the AI-assisted stop generation and validation pipeline
• Integrating geographic validation with Mapbox and OpenStreetMap
• Testing LLM failure, fallback and geospatial validation paths
• Developing subscription, billing and Premium tier enforcement
• Contributing to Docker, Gunicorn, Cloud Run and Google Cloud Storage deployment
• Leading product delivery through to pilot testing with 41 users — 15 guides and 26 tourists

View repository → View case study →

A human-in-the-loop AI assistant designed to help small businesses respond to customer enquiries without giving an AI system uncontrolled access to real conversations.

FastAPI · Python · PostgreSQL · pgvector · Gemini · RAG · WhatsApp Cloud API · Telegram · Redis · Docker · React · TypeScript

The system:

Customer message → Classification → Knowledge retrieval → AI draft → Human approval → Customer reply

I designed and built the MVP end-to-end, including:

• WhatsApp webhook ingestion and outbound messaging
• RAG over business-specific knowledge using PostgreSQL + pgvector
• Gemini-based classification and response generation
• Telegram Accept / Edit / Reject approval workflow
• Multi-tenant administration interface
• Redis-backed rate limiting
• Dockerised local infrastructure
• Authentication and safeguards around AI-powered endpoints

The key design principle was simple: AI can draft, but a human remains responsible for what reaches the customer.

View case study →

Backend Python · Django · FastAPI · Java · Spring Boot

AI & Data Gemini · LangGraph · RAG · pgvector · PostGIS · OR-Tools

Infrastructure PostgreSQL · Redis · Docker · Google Cloud Run · Google Cloud Storage

Frontend & Integrations React · TypeScript · JavaScript · REST APIs · Stripe · WhatsApp Cloud API · Telegram Bot API

I’m particularly interested in the engineering around the “clever” part of a system:

• making AI output verifiable rather than blindly trusted
• designing clear boundaries between services
• handling failures and degraded states intentionally
• turning product requirements into independently deliverable engineering work
• making local development and deployment reproducible
• building software around the way people actually use it

🌐 maxcorti.github.io

You can find deeper project breakdowns, architecture, contributions and lessons learned there.

LinkedIn → Email →

_{Some historical AURA contributions and PR merges appear under my previous GitHub account, MaxCorti1.}