> Hi, I'm Antonino Manuele Gargiulo

Sales teams at DGS build a profit-and-loss statement (Conto Economico) for every offer — costs, resources, margins — to check it's viable before it goes out. It was compiled by hand in a massive Excel template: no versioning, no access control, easy to get wrong. I built a webapp replacement where cost and resource data sync automatically from CRM (Dynamics 365 Sales) and ERP (F&O) instead of manual entry, role-based access ensures each manager only sees the offers they own, and a Hono.js + Drizzle ORM backend on PostgreSQL keeps the calculations consistent. React frontend with shadcn/ui and TanStack Query. Deployed on Azure AD for auth.

Postal analysts were bottlenecked by SQL expertise — every data question meant a ticket to engineering. I built a natural-language chatbot that lets business users query parcel delivery data directly, powered by a ReAct agentic pipeline over AWS Athena and an S3/Glue data lake. The full AWS infrastructure — ECS Fargate microservices behind an ALB, DynamoDB, CloudFront, Cloud Map for service discovery — was written as code with Pulumi and deployed via OIDC-authenticated GitHub Actions.

Doctors at one of Italy's top hospitals needed a unified interface across 15+ specialist systems to support stroke diagnosis. I built the React frontend and integrated an AI stroke-prediction engine (XGBoost/SHAP) that surfaces risk scores in real time alongside patient data. A Django + LDAP auth service handles identity across the microservices, orchestrated with Docker Compose and Traefik, shipped via GitLab CI/CD.

Aerospace engineers were chaining Nastran, Matlab, and HyperMesh simulations through ad-hoc scripts and manual handoffs. I built a visual workflow editor where they can compose and run simulation pipelines with drag-and-drop. React + React Flow handles the canvas; a Hono.js backend with Drizzle ORM manages execution state. Deployed on Kubernetes via GitLab CI/CD.

A media production company was spending hours transcribing meetings and manually assembling production briefs. I built a platform that ingests recorded meetings via Amazon Transcribe, extracts action items, and feeds a generative AI pipeline to produce documents and visual assets automatically. Infrastructure — CloudFront + S3 frontend, EC2 backend, SQS + Celery async workers, PostgreSQL on RDS — is managed end-to-end with Pulumi.

Mastercard's tourism analytics product needed its frontend rebuilt to keep pace with its data volume and analyst expectations. I led the UI/UX redesign with Next.js 14, building a range of chart types with Recharts to surface spending data across Italy's regions. The backend runs on Amazon EC2 with automated CI/CD via GitHub Actions.