Alberto Tuveri

Alberto Tuveri

Software & AI EngineerCreative Lead

From database schema to real-time WebGPU rendering and motion design.

Now, since May 2026

Now atSerSan

I own the front-end, WebGPU/3D and motion layer of what the studio ships, the AI work on its flagship platform, and the studio's creative direction.

SerSan is a custom software, automation and AI studio: Sersan Limited, London. I work remote, as Software & AI Engineer and Creative Lead, and I built the studio's own multi-agent system, SerSan Agent OS.

sersan.io

The sersan.io home hero at rest: the hexagonal SerSan mark and the glowing SERSAN wordmark, both drawn in particles, above a raymarched black hole edged by a thin bright ring, on a dark starfield.

Eleven works

At SerSan, the multi-agent system I built for the studio, a trading platform and its AI mentor, a workshop ERP, a WebGPU studio site, a real-estate platform and a brand film. Then Obla, my personal project, and three projects from outside SerSan.

  1. 01 / 11

    SerSan · multi-agent system

    SerSan Agent OS

    Agents with roles, hand-offs under a brake and a Gate on every action that goes out; 2,513 tests passed.

    Open the case

    Earned autonomy

    each agent acts on its own only at the level it has earned

  2. 02 / 11

    SerSan · live at spherenode.com

    SphereNode

    Vertical SaaS that replaced eight tools for the Italian retail-trading market.

    Open the case

    The spherenode.com home page, dark with a green accent.
  3. 03 / 11

    SerSan · RAG mentor

    Sphere AI

    The bilingual mentor inside SphereNode, measured on a 161-question evaluation set.

    Open the case

    A Sphere AI conversation: a trader's question, the answer with numbered citation markers, and the lessons it drew on listed as source cards, their titles redacted.
  4. 04 / 11

    SerSan · workshop ERP

    Maintech OMA

    Job orders on a six-phase state machine, ten trade portals.

    Open the case

    9

    roles, default-deny on every table

  5. 05 / 11

    SerSan · Jun – Sep 2026

    sersan.io

    Seventeen scroll-locked scenes on one continuous WebGPU canvas.

    Open the case

    The sersan.io intro: the SerSan mark built of blue particles above the glowing wordmark, both over the bright, bent rim of a black hole.
  6. 06 / 11

    SerSan · real-estate brand platform

    Domus Tua

    Mobile LCP 4.75 s → 1.74 s on real phones, no layout shift.

    Open the case

    The method section of the Domus Tua site, in ivory and red: a serif heading above the first steps of the method, set as cards along a winding red line.
  7. 07 / 11

    SerSan · identity and film

    Brand film

    Hexagonal mark, custom display typeface, colour system, and the brand film.

    Open the case

    A frame of the SerSan brand film: a glass browser card on blue, the glowing SerSan mark and wordmark above a black hole on it.
  8. 08 / 11

    Personal project · multi-agent system

    Obla

    One chief of staff I talk to, and the agents it hands the work to, on my own machine.

    Open its page

    Personal

    my own multi-agent system, for me alone

  9. 09 / 11

    ALS MCL Civitanova · freelance

    Badante24h

    Sub-second caregiver search; zero known auth vulnerabilities at launch.

    Open the case

    Sub-second

    geospatial caregiver search, PostGIS 3

  10. 10 / 11

    DOIT S.r.l · internship

    Voice AI Agent

    Meetings and Dynamics 365 contacts, from natural-language speech.

    Open the case

    Hands-free

    appointments in Microsoft Teams, from audio notes

  11. 11 / 11

    University of Camerino

    Supply Chain

    Ten roles, a team of three, on Spring Boot with JWT RBAC.

    Open the case

    The team's class diagram of the supply-chain platform, redrawn in light lines on a dark ground.

Measured before and after.

Two sets of figures with a before and an after: a 161-question bilingual evaluation set for Sphere AI, and page speed on real phones for Domus Tua.

Sphere AI · SerSan

Answer reliability, before and after

On a 161-question bilingual evaluation set. Citation correctness 100%, with zero locked-content leaks and zero fabricated citations.

  1. Answer-policy accuracy

    52%to68%

  2. Over-answering on questions to withholdLower is better

    56%to16%

  3. Clarification precision

    75%to94%

  4. Clarification recall

    32%to90%

  5. Unsupported claimsLower is better

    1.9%to0.6%

    Too small to draw on this scale.

Domus Tua · real phones

Fast on the phone in your hand

The mobile delay was traced to the cookie banner and the intro rebuilt to run in CSS rather than JavaScript. The cinematic experience stayed, with no layout shift.

  1. Mobile Largest Contentful PaintLower is better

    4.75 sto1.74 s

  2. Mobile blocking timeLower is better

    572 msto274 ms

Capabilities

Eight groups of tools, each tied to the work that proves it.

  1. 01

    Core

    • TypeScript
    • JavaScript (ES2024)
    • Python
    • Java
    • C / C++
    • C#

    Proof Supply chain

  2. 02

    Front-End

    • React 19
    • Next.js 16 (App Router)
    • Angular
    • Tailwind CSS v4
    • GSAP
    • Lenis
    • Framer Motion

    Proof Domus Tua

  3. 03

    3D & Graphics

    • WebGPU
    • WGSL / TSL
    • three.js
    • React Three Fiber
    • GLSL / WebGL2
    • GPGPU compute
    • Blender

    Proof sersan.io

  4. 04

    Back-End & DB

    • Node.js
    • Deno
    • Spring
    • Supabase
    • PostgreSQL
    • PostGIS
    • pgvector
    • MySQL

    Proof Maintech OMA

  5. 05

    AI

    • Multi-agent systems
    • Claude Agent SDK
    • Claude Code / MCP
    • Anthropic / OpenAI / Gemini
    • Local models (llama.cpp)
    • RAG & embeddings
    • LLM evaluation & guardrails
    • Copilot Studio

    Proof SerSan Agent OS Obla Sphere AI

  6. 06

    Cloud & DevOps

    • Vercel
    • Azure (AD, Speech)
    • Cloudflare (Stream / R2)
    • GitHub Actions
    • Sentry
    • Docker

    Proof SphereNode

  7. 07

    Testing & Tooling

    • Playwright
    • Vitest
    • pgTAP
    • Lighthouse CI
    • ESLint
    • Prettier
    • Git

    Proof Maintech OMA

  8. 08

    Creative

    • Motion graphics & VFX (After Effects)
    • Video editing & post-production
    • Brand identity & logo design
    • Art direction
    • Typography
    • Sound design

    Proof Brand film

Camerino,Trento,London

A B.Sc. at Camerino, an M.Sc. in AI Systems at Trento, and professional work since November 2025, remote for a London studio since May 2026.

  1. Sep 2026 – Present Education

    M.Sc. in Artificial Intelligence Systems

    University of Trento, Trento, Italy

  2. May 2026 – Present Experience

    Software & AI Engineer, Creative Lead

    SerSan, AI studio · Sersan Limited (London) · Remote

  3. Jan 2026 – Present Experience

    Full-Stack Developer

    ALS MCL Civitanova · Badante24h, a production PWA · Freelance · Marche, Italy

  4. Nov 2025 – Feb 2026 Experience

    AI & Software Developer

    DOIT S.r.l · a voice AI agent for Microsoft Teams · Internship · Fabriano, Italy

  5. Jan 2022 – Apr 2026 Education

    B.Sc. in Computer Science

    University of Camerino, Camerino, Italy

    Thesis: From Autocomplete to Agentic Orchestration: Architectural Analysis and Security of OS-Level Agents

Four waysto work with me

I help companies and private individuals put today's AI tools to work, from a first video to a team of agents that runs on its own. Each engagement starts from what you need to deliver, not from the tool.

  1. 01

    AI content generation

    Images, video and copy made with generative models and directed like any other production: a brief, a look, a review, a delivery.

    Proof On request

  2. 02

    Video editing

    Editing, motion graphics, colour and sound design for brand films, product videos and social cuts.

    Proof Brand film

  3. 03

    Claude Code best practice

    A setup that works, for you or your team: project instructions, skills, hooks, MCP servers and review gates, so the agent follows your rules instead of guessing.

    Proof Obla

  4. 04

    Multi-agent systems, built and tested

    Systems where several agents share a job: each agent with its role, hand-offs with a brake that calls in a person, a gate on every action that goes out, and tests that show what happens when a part fails. I built one for SerSan, and Obla as my personal project.

    Proof SerSan Agent OS Obla

For companies and for private individuals.

Write to me On to Obla

Obla

Obla is my personal multi-agent system: one chief of staff I talk to, and the agents it hands the work to, instead of opening sessions, writing prompts and checking on things by hand. It lives on my desktop, runs on my own machine and answers to me alone.

Personal project

  1. 01

    It listens

    I say its name, or hold a key, and speak. The wake word and the voice chain run locally.

  2. 02

    It thinks

    A small local model answers closed questions, such as which agent takes a job; a second one holds the conversation; only real work goes to Claude Code.

  3. 03

    It works

    Each job goes to the agent that fits it, for code, design, research, content or review: a Claude Code session on the right repository, with the brief already written. The agents hand work on to each other, within limits written in code.

  4. 04

    It asks

    Models propose, code acts: a push, a merge or a deletion waits for my yes on an approval card, and a message to someone else stays a draft until I send it.

  5. 05

    It reports

    Every morning at 08:00: jobs, sessions, approvals, spending and what each agent did, spoken or written. What it remembers is kept in Markdown notes I can read and edit.

Runs on
Windows, TypeScript on Node.js, Rust and Tauri, llama.cpp with local models, the Claude Agent SDK and MCP, SQLite, Obsidian
For
One person. Not distributed.

Write to me.

For a role, a project, or one of the four ways to work together.

albertotuveri@gmail.com

LinkedInGitHub