Custom software design and development
I build web apps, internal tools, and integrations around your operational problem, not around a technology.
Independent software engineer · Brescia
Software and AI systems, taken to production
Gianluca Mazza is an independent software engineer who builds production AI systems for technical teams and technical founders shipping LLM-backed systems — durable state, deterministic recovery, reproducible evals, and cost under control.
I work in short delivery cycles you judge by using, hand over code and documentation you own, and add AI only where it earns its place — with human oversight, audit trails, and cost under control.
Summary: I build AI systems that keep running after the demo, and I hand over the code that does it. My prior work spans ten years on distributed financial systems: Bitcoin wallets, Lightning prototypes, Liquid integrations, and bank custody systems.
I hand over code the client owns: repository, documentation, and access. No dependency on me. Published software, citable via DOI: 10.5281/zenodo.19972258
What I do
Three ways I help technical teams and technical founders get software built around how they actually work.
Technical coordination and fractional CTO
I provide part-time technical leadership when you work with vendors or a small team: stack choices, quote and code review, roadmap.
AI integration and process automation
I add AI where it brings verifiable value, with human oversight and costs kept under control.
What I have built
A look at systems I have actually shipped and the concrete problem each one solves.
emotional-memory
#emotional-memory
Problem
When a business adds an AI assistant, how do you verify that what it remembers today will still recall correctly after the next model or software update?
Outcome
A published package and reproducible benchmarks give you inspectable evidence before trusting an AI memory layer in production.
orka
#orka
Problem
Requests arrive from chat, email, and internal tools, but without a single durable queue there is no reliable path from channel input to a tracked, reviewable LLM workflow.
Outcome
One prioritized queue routes multi-channel requests into bounded LLM workflows with MCP/A2A support.
OpenFatture
#openfatture
Problem
How do you ship FatturaPA/SDI-compliant invoicing with AI assistance without handing business data to another SaaS subscription?
Outcome
Compliant invoices produced and sent without a SaaS subscription: business data stays local and every external effect passes manual review.
Experience
Co-Founder @ Venere Labs
September 2024 - PresentItaly
I co-founded an independent AI engineering lab, working on inspectable agent systems and reproducible evals.
LangChainLangGraphMCPOpenAI/Anthropic
AI Director, contract @ Kiara Industries
August 2025 - November 2025Brescia, Italy
I led a contract AI product build from requirements to release, covering architecture, implementation, and technical delivery.
Product DesignLLM SystemsTechnical Delivery
Co-Founder and CTO @ Scaling Parrots
April 2020 - November 2025Brescia, Italy
I led technical direction across product and infrastructure, turning early product decisions into shipped systems.
BitcoinPythonRustAWS
Bitcoin Developer @ Eidoo
2018 - August 2019Ticino, Switzerland
I shipped Bitcoin wallet features and explored Lightning, Liquid, and key-management architecture.
Bitcoin WalletsLightning NetworkLiquid NetworkCryptography
Technical Operations Lead and Co-Founder @ 0301 Mining Farm
July 2017 - April 2019Sofia, Bulgaria
I ran a Bitcoin mining operation in Bulgaria, covering infrastructure, hardware efficiency, and procurement.
Bitcoin MiningHardware OptimizationOperations Management
Senior Software Developer @ BHB - Blockchainlab
January 2017 - January 2019Milan, Italy
I built Bitcoin and blockchain systems for clients, from prototypes and security protocols to technical conference material.
BitcoinBlockchainPoC DevelopmentSecurity Protocols
Common questions
- Who do you build software for?
- Technical teams and technical founders shipping LLM-backed systems that must survive contact with production. Usually the prototype already works and nobody owns the job of making it hold up against real traffic, real data, and a real budget.
- Can you work with an existing team and stack?
- Yes. I plug into the stack and constraints you already have. Typical work: hardening an LLM workflow with durable state and deterministic recovery, putting evals around a system before it ships, or bounding its cost — and I leave your team documentation and tests it owns.
- What proof do you have that your systems work?
- My work is open source and verifiable. The emotional-memory project has a Zenodo DOI (https://doi.org/10.5281/zenodo.19972258) with reproducible benchmarks; systems like orka and OpenFatture are public on GitHub. I show qualitative outcomes and real artifacts, not invented numbers.
- How do I start a project with you?
- Write me a few lines about the operational problem and the systems involved — that is enough for a first assessment. I reply by email; there is no form or sales call to book.
Have a project in mind?
Tell me about the operational problem you want to solve. A few lines about the context and the systems involved are enough for a first assessment.
Technical deep dives
System design notes, the architecture of this site, and public code — the detail behind the claims on this page.