Building
quiet systems,
carefully.
I'm Lorenzo. Eight years of C#/.NET across backend architecture, workflow automation, AI integration, and Unity — for studios, banks, and ambitious teams.
Work
-
001 Everywhere srl Lead software engineer & architect The distributed workflow engine I built and now lead — processing over a million sessions a year, automating ~80% of operator work with OpenAI, and shipping every week at 99.9% uptime. 2017—now I'm the lead engineer and architect on Everywhere — an enterprise workflow automation platform unifying business communications across email, SMS, WhatsApp, voice, and Slack. I own the technical direction, the architecture, and the team of five I work with day to day.
- I architected the distributed workflow engine that orchestrates cross-channel business automation — processing 1M+ sessions a year across enterprise scenarios.
- I built and maintain the fault-tolerant, real-time backend services on Azure that sustain 400+ concurrent sessions at 99.9% uptime in production.
- I integrated OpenAI and ML models into core business logic, automating ~80% of process creation and testing that previously required manual operator work.
- I lead a cross-functional team of 5 (developers, designers, marketing, QA), coordinating delivery with stakeholders and owning end-to-end technical direction.
- I designed the backend that unifies email, SMS, WhatsApp, voice, and Slack into a single workflow layer — replacing previously siloed operational channels.
- I integrated Stripe payments and Italian SDI e-invoicing into the automated workflows, enabling compliant end-to-end billing.
- I modernised the release process from monthly to weekly deployments while improving reliability and reducing rollback incidents.
Blazor .NET C# Azure MySQL SignalR Microservices Real-time systems OpenAI SDK Machine learning Stripe Payment APIs Workflow automation System architecture -
002 MAJC Lead full-stack engineer I led the architecture for an AI-powered hospitality workforce platform — designed the identity, monetisation, and subscription foundation the downstream teams now build against. 2025—2026 I came in as the architecture lead on MAJC, an AI-powered hospitality workforce platform unifying job matching, community, and training. My job: audit what existed, design the foundation, and write the ADRs that let the rest of the team execute against a coherent plan.
- I led the architecture for the platform, unifying Jobs, Community, and Academy into a single product ecosystem with a shared identity layer.
- I designed the identity and monetisation foundation by integrating Azure AD B2C, Stripe, and Microsoft Fabric to support multi-tier subscriptions across the platform.
- I audited the existing platform and delivered a scalable technical roadmap for authentication, role-based access, and subscription logic across services.
- I structured the technical foundation for a multi-vertical product where Jobs, Community, and Academy share a unified identity and subscription layer.
- I authored architecture decision records and technical documentation enabling downstream teams to execute against a clear, scalable foundation.
.NET C# Azure Azure AD B2C Microsoft Fabric Stripe API Microservices Architecture Documentation Technical leadership -
003 Qualifirst Foods Full-stack AI engineer I designed and built a Slack-integrated GPT-4o knowledge bot end-to-end — RAG over FAISS, FastAPI back-end, scheduled syncs — collapsing GitHub, Drive, Snowflake, and Postgres into one conversational interface. 2025 I owned the design and build of Qualifirst's internal AI knowledge bot — backend, RAG pipeline, integrations, deployment. The brief: pull fragmented knowledge across many tools into one place the team already lives in (Slack).
- I built a Slack-integrated AI knowledge bot using OpenAI GPT-4o, unifying GitHub, Google Drive, Snowflake, and PostgreSQL into a single conversational interface for internal teams.
- I designed the RAG pipeline with FAISS vector search over internal documentation, enabling semantic retrieval across heterogeneous sources (repos, docs, structured data).
- I architected a FastAPI back-end exposing real-time endpoints for GitHub issues, Drive files, and Snowflake queries — secured with API key auth and least-privilege scopes.
- I implemented APScheduler-based background sync jobs maintaining a local cache across GitHub, Drive, and Snowflake — reducing external API calls and improving response latency.
- I migrated the system from a REST/sync-based architecture to a Slack + OpenAI bot interface, simplifying access and centralising knowledge inside existing team workflows.
- I integrated GitHub, Google Drive, Snowflake, PostgreSQL, Slack, Odoo, and Read.AI under a unified authentication and data-access layer with secure credential management.
Python FastAPI OpenAI SDK LangChain FAISS RAG MCP Snowflake PostgreSQL Slack API GitHub API Google Drive API Odoo Microservices -
004 Research Theory Unity game developer I built a Unity 2D cognitive-assessment game from the ground up — four psychometric tasks, semantic-vector scoring, A* procedural maps — and compressed an 8GB embedded model to 70MB to fit a WebGL build. 2024—2025 I built the entire game — engine architecture, the four psychometric tasks, the scoring engines, the data export pipeline. The product is a Unity 2D cognitive-assessment game asking whether gameplay performance can estimate a player's IQ, deployed in WebGL for accessible research participation.
- I built the Unity 2D cognitive-assessment game from the ground up, deployed in WebGL for accessible research participation.
- I compressed an embedded AI model from 8GB to 70MB — a 99% reduction — to fit WebGL delivery constraints while preserving inference quality. The most amazing thing I've shipped.
- I implemented four distinct psychometric tasks — RAT (verbal convergent), DAT (verbal divergent), Mirror (nonverbal divergent), Raven (nonverbal convergent) — each with custom scoring algorithms.
- I built the DAT scoring engine using cosine-distance similarity over semantic word vectors loaded from compressed binary data, computing divergence scores from player input in real time.
- I developed Perlin-noise procedural map generation with A* pathfinding, including a villager-simulation system that validated maze solvability under randomised seeds.
- I architected the codebase using a Manager-Service pattern with event-driven communication, ScriptableObject-based configuration, and SOLID principles for extensibility.
- I built Firebase-backed data export pipelines capturing gameplay sessions, minigame responses, and final scores for downstream research analysis.
Unity Unity3D C# .NET 8 WebGL Firebase Procedural generation A* pathfinding Semantic vectors ScriptableObjects Event-driven architecture 2D Games Cognitive assessment -
005 IC Group Unity architect consultant I came in as the senior architecture and security consultant — reviewed the codebase, wrote the standards, and integrated an OWASP-aligned process into the team's Azure DevOps workflow so they could keep the bar high after I left. 2024 I was the senior architecture and security consultant for a Unity puzzle-game studio looking to lock in code quality and security standards before scaling. The deliverable: not just a review, but a repeatable process the team owns going forward.
- I conducted in-depth code and architecture reviews of the Unity puzzle-game codebase, identifying scalability bottlenecks and recommending structural improvements for long-term maintainability.
- I authored coding standards and review-workflow documentation, establishing a repeatable process the team uses to maintain code quality across future development.
- I evaluated the project against OWASP-aligned security requirements, flagging vulnerabilities and recommending remediation within the existing QA and deployment pipeline.
- I defined secure development guidelines covering input validation, secret management, and dependency hygiene, integrated into the team's Azure DevOps workflow.
- I delivered structured technical documentation enabling internal engineers to apply review standards and security practices independently after the engagement.
Unity C# .NET 2D Games OWASP Secure development Azure DevOps Code review Software architecture Documentation -
006 Fogale LensX Unity engineer I built a Unity-on-iOS proof of concept for Fogale's patented optics technology — bridged native C++ image-processing into the Unity runtime to hit demo-grade frame rates on consumer iPhones. 2024 I built the proof of concept end-to-end. Fogale is commercialising patented optics technology and needed a presentation-quality mobile demo to put in front of prospective manufacturing partners — Unity on iOS, with native C++ doing the heavy image-processing work.
- I built the Unity-based mobile proof of concept demonstrating the patented optics technology, delivering presentation-quality output for stakeholder demos.
- I integrated C++ image-processing routines with Unity on iOS, bridging native performance-critical code with the Unity runtime for real-time mobile execution.
- I optimised the mobile image-processing pipeline to sustain demo-grade frame rates on consumer iOS hardware, balancing visual fidelity against real-time performance constraints.
- I implemented mobile-side functionality with attention to device variability, memory constraints, and hardware-acceleration limits typical of consumer iOS devices.
- I delivered a presentation-ready prototype used in pitches to prospective manufacturing partners, supporting commercialisation of the underlying patented technology.
Unity C# .NET C++ iOS Image processing Performance optimisation XR Mobile -
007 Henry Schein Unity engineer A year-long engagement building an iOS and Android AR application for clinicians — real-time face tracking, OAuth into internal systems, shipped into a regulated healthcare context. 2022—2023 I led the Unity / AR engineering on a year-long production engagement for Henry Schein, the global healthcare distributor. The product simulates post-treatment dental outcomes — clinicians use it to show patients cosmetic results before procedures.
- I built the iOS and Android AR application simulating post-treatment dental outcomes, enabling clinicians to visualise cosmetic results with patients in real time before procedures.
- I implemented real-time face reconstruction and tracking using AR Foundation (ARKit / ARCore), aligning 3D dental overlays to facial geometry across varying lighting and head poses.
- I optimised rendering and AR pipelines to maintain smooth performance on consumer iOS and Android hardware, balancing visual fidelity with mobile resource constraints.
- I integrated the application with Henry Schein's internal systems via OAuth, supporting authenticated clinician workflows and production deployment in a regulated healthcare context.
- I delivered a production AR product on a year-long engagement, navigating healthcare data-handling requirements and cross-platform release cycles.
Unity C# .NET AR Foundation ARKit ARCore Face tracking iOS Android OAuth JavaScript -
008 Aescape Unity engineer I owned the real-time anatomy visualisation system for an AI-driven robotic massage platform — Blender-to-Unity asset pipeline, live mesh deformation, and the engineering tools used to validate massage paths. 2022 I owned the real-time anatomy visualisation for Aescape's AI-driven robotic massage platform — from the Blender-to-Unity pipeline through the live mesh deformation to the engineering tooling. The system links live body-scan data to a procedural muscle and skeletal model in real time.
- I built Unity-based real-time anatomy visualisation for the AI-driven robotic massage platform, linking live body-scan data to a procedural muscle and skeletal model.
- I implemented high-performance mesh updates and proportional muscle deformation during dynamic pose changes, sustaining real-time frame rates under continuous body-scan input.
- I designed a procedural anatomy system mapping live body-scan geometry onto a rigged character model, accommodating individual variation in proportion and posture.
- I created interactive visualisation tools for muscle regions and motion trajectories, supporting engineering review and validation of robotic massage paths.
- I built a Blender-to-Unity asset pipeline for rigged anatomy models, including custom skinning and bone-weight workflows tailored to real-time deformation requirements.
Unity C# .NET Blender Mesh deformation Real-time rendering Shaders Character rigging Body scan data 3D animation -
009 AMIDEAST Full-stack engineer I led the migration of a legacy VB6 / VB.NET reporting portal to .NET Core + Blazor across MENA region offices — phased rollout, schema rebuild, feature parity preserved. 2021—2022 I led the migration end-to-end. AMIDEAST runs educational programs across MENA region offices on a VB6/VB.NET reporting portal carrying decades of accumulated technical debt. My job was to retire it cleanly onto a modern .NET Core + Blazor stack without losing a single report.
- I led the migration of the legacy VB6 / VB.NET reporting portal to a modern .NET Core + Blazor architecture, eliminating decades of accumulated technical debt.
- I rebuilt the SQL Server database schema and application logic, improving maintainability and enabling long-term scalability of the reporting platform.
- I optimised SQL Server queries and rebuilt UI rendering paths in Blazor, improving report-generation responsiveness across the portal.
- I delivered new reporting features alongside the modernisation, expanding analytical capabilities while preserving feature parity with the legacy system.
- I executed a phased migration path from the legacy system to the new platform, minimising downtime and ensuring continuity of reporting workflows for end users across MENA region offices.
.NET Core Blazor C# SQL Server VB.NET VB6 JavaScript TypeScript Azure DevOps MySQL -
010 Tecso Machine learning engineer I designed and built a data-agnostic ML system for insurance risk assessment end-to-end — TensorFlow models, Python pipelines, web-scraped public data, and classification decisions inside production. 2021 I designed and built the system end-to-end. Tecso wanted to automate insurance risk assessment with ML — and crucially, wanted a system that could ingest heterogeneous data without rewriting integration code each time a new dataset showed up.
- I built a data-agnostic ML system for insurance risk assessment, capable of ingesting heterogeneous dataset formats and producing automated classification outputs.
- I integrated TensorFlow models into Python-based data pipelines, enabling scalable analysis across structured and semi-structured insurance data sources.
- I designed the system to be dataset-agnostic, allowing onboarding of new data sources without retraining infrastructure or rewriting integration code.
- I improved insurance risk-evaluation workflows by replacing manual classification steps with model-driven decisions inside production pipelines.
- I built web-scraping components to enrich risk-assessment datasets with external public data sources, expanding the model's input coverage.
Python TensorFlow Machine learning Feature engineering Web scraping REST APIs MySQL SQL Risk analysis -
011 Atum Full-stack engineer I worked full-stack on a regulated insurance and private-banking platform — owning the wire-transfer integration with a major Italian bank, designing schemas and Swagger-documented APIs. 2020—2021 I worked across backend and frontend on Atum's regulated insurance and private-banking management platform — a long-lived monolithic codebase serving institutional clients. I owned the wire-transfer integration with a major Italian banking group, where compliance requirements drive every design decision.
- I developed backend features for a regulated insurance and private-banking management platform, working inside a long-lived monolithic codebase serving institutional clients.
- I integrated wire-transfer capabilities with a major Italian banking group, translating compliance requirements into production backend flows for inter-bank transactions.
- I collaborated directly with private-banking and insurance stakeholders to translate complex domain requirements into production-ready backend logic.
- I designed SQL schemas, stored procedures, and REST APIs (documented via Swagger) supporting business-critical workflows across the platform.
- I contributed front-end work in Angular and React on customer-facing modules, bridging backend services with end-user workflows.
C# .NET SQL Stored procedures Swagger REST APIs Angular React JavaScript Azure PostgreSQL MySQL Payment APIs Private banking -
012 AIV Unity3D teacher I designed and taught a Unity 2D course at AIV, the Italian academy for game development — OOP C#, scene management, UI, the full game lifecycle, mentoring students from concept to playable build. 2020 I designed and delivered the Unity 2D course at AIV, the Italian academy specialising in video game development. The aim was to build students' Unity engineering fundamentals progressively — concept to playable build — with structured one-on-one mentoring on architecture and gameplay implementation.
- I designed and delivered a Unity 2D development course covering object-oriented C# scripting, scene management, UI systems, and the full game lifecycle.
- I mentored students through complete project cycles from concept to playable build, providing one-on-one feedback on code architecture and gameplay implementation.
- I structured lessons and practical exercises to build Unity engineering fundamentals progressively, balancing conceptual theory with hands-on project execution.
- I taught at AIV — an Italian academy specialising in video game development — contributing to the training of next-generation Unity developers entering the industry.
Unity C# OOP Game development Technical teaching
Projects
- α BlazorDynamicForm Form generator for Blazor My most popular open-source library — 35+ GitHub stars, distributed on NuGet. Generates fully validated Blazor forms automatically from any C# class. Two lines of code replace a full page of markup; supports nested objects, lists, dropdowns, file uploads, and an embedded code editor. 2023
- β GraphCSharp Code graph GitHub Action A reusable GitHub Action that generates a live, interactive map of any .NET repository's internal structure on every push, publishing it as part of the documentation site. Always-current architecture diagram, zero manual effort. Supports zoom, filtering, and navigation. Live demo at thesmallpixel.github.io/GraphCSharp. 2025
- γ MultiClassTrainerML ML.NET train-and-serve A reference implementation showing how to take a machine-learning model from raw business data to a working API entirely in C# — no leaving the .NET ecosystem. A training pipeline pulls labelled data, trains with cross-validation, and saves the model; a Web API loads it and serves predictions through a versioned endpoint. 2023
- δ TwilioCallCenter Programmable cloud call center A production-shaped reference showing how a single Twilio number can become a fully programmable cloud call center — inbound and outbound calls, IVR menus, live agent routing, call recording, two-way SMS, all from one .NET service. I also contributed fixes upstream to the official Twilio SDK. 2024
- ε ChatGPT4CodeHelper Code review CLI A small CLI that pipes the parts of code review I'd rather not do through GPT-4 — diffs in, suggestions out. Used quietly across a few of my repos. 2023
- ζ GourceMultiRepo Repo-history visualisation A PowerShell script that stitches the history of many repositories into one continuous Gource visualisation. The kind of thing you write once for a talk and end up using every quarter. 2022
"The best system is the one you didn't have to explain."
— ll, on a tuesday
I write quiet, considered software.
I'm a senior full-stack engineer based in Milan with eight years in C#/.NET. My focus is backend architecture, workflow automation, and AI integration — and I'm drawn to hard problems: regulated banking systems, real-time platforms, embedded ML under tight constraints, legacy modernisation.
I lead engineering at Everywhere srl, where I architected a distributed workflow engine processing over a million sessions a year. Alongside that, I take on senior contracts through Toptal — recently architecture for an AI hospitality platform, a GPT-4o knowledge bot for an internal team, and a Unity cognitive-assessment game where I compressed an 8GB model down to 70MB to fit a WebGL build.
The most amazing thing I've shipped is probably that 99% model reduction — but the work I'm proudest of is quieter: turning a fragmented operational stack into a single workflow layer, moving a team from monthly to weekly releases, writing the kind of system you don't have to explain.
Want to work
together?
I'm taking on senior contracts. Backend architecture, workflow automation, AI integration, .NET services, Unity. Tell me what you're working on.
Or write directly to thesmallpixel@gmail.com. I usually reply within a couple of days.