$ whoami

Gianluca Sestito

Role: Lead Backend Engineer

Base: Milan, IT

I keep systems predictable when the things they depend on are not — webhooks that answer when they feel like it, and language models that are wrong often enough to need measuring. Currently: a multi-tenant generative platform in production, and commerce content across 67 locales.

67 Locales in Production

74% Energy Cut, Verified

6 Engineers Led

Let's Build → View Work

Featured Work

Generative AI Edge Platform

Multi-Tenant Generative Platform

External advertisers generate campaign creative that ends up on public billboards. Cloudflare Workers: isolated tenants, authorisation on every route, REST APIs on D1. Model agents and image generation chained on Trigger.dev — a generation takes minutes, so status streams back in realtime rather than timing out the request.

74% less display energy, verified independently under IPMVP
Eval datasets, LLM-as-a-judge scoring, output schema validation
Launched publicly across three European countries

Cloudflare Workers D1 Trigger.dev Langfuse TypeScript

Client platform · built and owned end to end

Webhook-Driven Composable Commerce

Headless Commerce Replatforming

Async middleware built from scratch to move an enterprise storefront off a monolithic suite onto headless. A gateway takes webhooks from CRM, OMS, PIM, CMS and search; Redis covers caching and the fast lookups. No broker in the stack, so a broker's ordering and delivery guarantees live in application code — across six systems I don't control.

Peak load spanning two orders of magnitude against low season
Idempotent jobs, exponential backoff, transient vs. real error taxonomy
Two peak seasons without customer-facing incidents

Nitro Node.js Trigger.dev Redis Datadog Sentry

Client platform · on-call from go-live

Side projects on GitHub: CV AI Enhancer ↗ Resilient Order Saga ↗

Experience

Alkemy+

Current

2023 — Present

Lead Backend Engineer · Backend Engineer 2023–24 · formerly Atoms

Platform Ownership at Scale

Two enterprise replatformings from scratch, live three months apart. Technical owner of one since go-live — 67 locales, several brands. I oversee most of what ships, run the reviews, scope and estimate the change requests.

                                67 locales
                                Production escalation
                            

Leading the Backend Line

Grew the line from 3 engineers to 6, five as direct reports. Wrote the practices from nothing: code review, Architecture Decision Records, branching and release rules, test-first by default.

                                3 → 6 engineers
                                4 mentored
                            

NTT DATA

2023

Backend Engineer

Regulated Entitlement Calculations

Extended a large legacy Java system calculating provincial welfare entitlements — a regulated domain with no tolerance for a wrong figure.

                                Legacy monolith
                                Regulated domain
                            

Legislation into Code

Turned changes in the law straight into calculation logic, validated against real data and shipped through controlled releases on a large monolith.

                                Real-data checks
                                Controlled releases
                            

GFT Technologies

2019 — 2023

Backend Engineer

Off Bare Metal onto GCP

Main contributor on moving an enterprise retail platform off bare metal onto Google Cloud: Dockerfiles, CI/CD pipelines on GitHub Actions, and how the services were containerised.

                                Bare metal → cloud
                                3.5 years of GCP
                            

Enterprise Backend Delivery

Feature work, maintenance and releases in PHP and Java (Spring Boot) for retail and other enterprise clients.

                                PHP · Spring Boot
                                Release management
                            

Skills

AI in Production

Agent orchestration, tool calling, RAG, embeddings, Vercel AI SDK, Gemini, OpenRouter

Langfuse tracing and prompt versioning, eval datasets, LLM-as-a-judge scoring, output guardrails. Measuring model quality rather than assuming it.

Core

Backend & APIs

Node.js, TypeScript, Nitro, Express, REST API design, Java (Spring Boot), PHP

APIs that stay predictable under load and legible to whoever extends them. TDD, code review standards, Architecture Decision Records.

Core

Cloud & Edge

Google Cloud, Cloud Run, Cloudflare (Workers, D1, Vectorize, AutoRAG), Docker, GitHub Actions

Three and a half years of daily GCP in production; the last year on the Cloudflare edge, with data isolated per tenant.

Proficient

Async & Integration

Webhook-driven orchestration, Trigger.dev, PostgreSQL, Redis, SQLite / D1, GraphQL

Idempotency, exponential backoff, error taxonomies separating transient failures from real ones — the delivery guarantees you write yourself when there's no broker.

Proficient

Composable Commerce

Commerce Layer, Amplience, Algolia, Auth0, Adyen, PayPal, Klarna, Yext

The integration surface of a headless storefront: content modelling, delivery pipelines, payments, search.

Reliability & On-Call

Datadog, Sentry, incident response, alert threshold and severity design

Formal on-call rotation and first escalation from go-live, with alert thresholds and severity levels agreed with the client's ops team.

Languages: Education:

Contact

Backend systems, AI in production, or a team that needs practices to grow into. Open to remote across the EU, and to relocation.

@ Email gianluca.sestito93@gmail.com → in LinkedIn Connect with me ↗ </> GitHub View my code ↗