I partner with business owners as a Fractional CTO: supervising external vendors, dismantling inflated quotes and lock-in contracts, and restoring architectural order to company systems. Operating margin grows when unnecessary daily friction is taken away from the people doing the work. I stand on the company’s side: zero conflict of interest, no retainers to push.
Full-stack developer, software architect, SEO & information retrieval
Managing technology solely through agency slide decks leads to overpaying tenfold; an isolated developer ignores business workflows. I unite executive financial oversight with hands-on code verification to protect margins and free operational capacity.
The vendor writes the quote, estimates the hours and picks the architecture: nobody inside can tell whether the job is worth two days or twenty. I read the contracts, decide what to buy and what to build, and manage developers one priority task at a time. Code the company paid for stays with the company.
I built ERPs single-handedly that have run on national networks for years, holding the history of hundreds of companies. That is how I know what a change really costs, and when a closed platform is holding your data hostage. Good back-office software spares people from retyping data from one screen to another.
A careless migration wipes out years of rankings overnight, and the next day leads are bought on Google Ads. I have studied search engines since 2006: historical URL redirects, crawlability and real browser speed are part of the project from the first draft.
A language model is good at reading a messy PDF or a confusing email, and bad at arithmetic. I use it to extract and classify data from documents; calculations, commissions and reconciliations stay in deterministic code, wired into the ERP and compliant with GDPR and the AI Act.
Four real contexts where software architecture, complex migrations and organization produced verifiable numbers. The Il Tuareg and OSM systems run without me: support is handled by dedicated, autonomous people, and the work is almost always new features, almost never bug fixes.
External Fractional CTO (collaboration since 2020, appointed in 2026)
I started working with them in 2020, the year tourism stopped, to rebuild the website. In 2021 the Estate INPSieme tender arrived with a mountain of applications: I wrote a custom module inside WordPress to handle them. It did the job, but it was slow and awkward, and the team had to bend its way of working to its screens. By 2023 Language Travel was over half of revenue and every application still went through it. In 2024 I rewrote it from scratch as a standalone CRM in Svelte and Firebase, starting from how operators actually work: the software adapts to the workflow, not the workflow to the software. In 2025 I added AI to read incoming documents. The result belongs to the team: Emanuela Lo Cascio and Filippo Falanga, who run Language Travel, were clients able to work as a team and keep communication clean. They built the market; the software let the division absorb the growth without adding people. Since Summer 2026 I have been the external Fractional CTO for the whole company.
Source: filed financial statements (VAT 04168670828). Revenue as an index, 2019 = 100: down 85% in 2020, 2.6 times pre-COVID by 2025. Net profit uses its own scale: read the shape of the curve.
Joined in 2020 (tourism COVID nadir) to rebuild the corporate website.
First custom WordPress module to handle INPSieme tender applications.
The back office leaves WordPress and is redesigned around the operators’ workflow: standalone Svelte + Firebase CRM, automated document handling.
From the Language Travel division alone to technology leadership across all of Il Tuareg.
Division share of total revenue, year by year, and the software it ran on.
IT Architect and Full-stack Developer of the ERP (solo)
I designed and built the ERP on my own, across frontend, backend and database. On the client side I worked closely with Gaetano Seminara: he knew what was needed and said it clearly, without creating confusion. The project ran in two phases:
Built to replace legacy software across a 20-partner franchise. Previously they ran 20 disconnected systems on 20 isolated databases with conflicting IDs and separate credentials, forcing manual spreadsheet compilation. Unified all 20 databases into a single centralized proprietary ERP.
Given the operational success at OSM Value, the system was scaled nationally for the entire OSM Group across hundreds of branches and companies, managing multi-tier commissions, Business Intelligence, and analytics.
The national group brought 4 different ERPs used by hundreds of companies. Years of invoices, quotes, deliveries, collections and stats moved into one system, with accounting balances verified on every migrated record.
Every new partner meant one more system to keep aligned by hand.
Designed and built single-handedly: one database, one account per person.
Scaled nationally across hundreds of branch enterprises: multi-tier commissions, BI, and proprietary architecture.
Commissions, rankings and stats come straight out of the ERP.
Fondatore & Lead Systems Architect (Web magazine founded in 2004)
A personal editorial publication I have operated for over 20 years. Rebuilt from scratch in the past year, migrating ~20,000 articles from monolithic WordPress to a fast Svelte + Firebase architecture with proprietary analytics designed without personal data collection and zero tracking cookies.
Real user assessment (CrUX): passed. Lab Speed Index: 0.5s · Total Blocking Time (TBT): 30ms.
Verify live report on Google PageSpeed ↗Official laboratory benchmark · High-performance Svelte systems
The field demonstration of my architectural philosophy: zero bloated frameworks, route-level code-splitting, inlined critical CSS, deferred scripts, and zero tracking cookies. Verified live by Google’s official auditor under emulated slow 4G mobile conditions.
Official Google PageSpeed Insights audit with Lighthouse 13.5.0 on emulated Moto G Power, single page session, and slow 4G throttling.
Verify live report on Google PageSpeed ↗Co-founder & Fullstack Architect (Concluded experience)
Multi-tenant corporate welfare platform built from scratch in Node.js and Svelte: wallet ledgers, cryptographic B2B/B2C voucher issuance, and automated accounting reconciliation.
The architecture allowed the startup to scale rapidly across multiple corporate welfare plans, delivering a key market advantage: allowing employees to spend their welfare credits with any merchant or provider chosen by the company or employee, avoiding rigid closed catalogs.
Paying skilled people to act as human photocopiers between spreadsheets and disconnected software burns valuable hours and slows the business down. Drag the three sliders to estimate the financial impact of manual friction on your team:
Capacity redirected to core tasks: Hours taken away from repetitive tasks go back to customers, errors drop, and people work better and are more likely to stay.
Zero hiring tax: With proprietary architecture, when your team doubles, license costs stay the same: you only pay the people maintaining the software. Under defined conditions, software qualifies as a capitalized asset on the balance sheet.
Original research essays and algorithmic analyses (2006–2009), rewritten and verified by Vincenzo Lombino through the lens of modern distributed systems and AI.
The blog's first post, September 2006: looking at search engines without siding with white or black hat.
A 2006 link experiment and what it still says about relevance.
September 2006: language, relevance and spam shift along with the words people use.
Mass-generated text fools filters only briefly. The 2006 Anti-Markov experiment, reread with today's language models in mind.
How a 2006 doorway page generator worked: synthetic text, keywords in tags and internal links.
Google's search operators one by one: boolean logic, URL, title and text filters, dates and wildcards.
The 2006 guide to regular expressions, with a new chapter: one badly written regex can stall a server.
Negative offsets, display:none and cloaking on User-Agent, Referer and IP: what the engine saw versus what the user saw.
January 2007: source code and the rendered page do not match, from static HTML to dynamic client-side DOM.
The January 2007 discussion with Agoago and Fradefra: what Googlebot really indexes, iframes and JavaScript, on-screen text selection.
March 2007: reading browsing history from the browser cache and profiling users without consent.
Artificial identities against reputation systems: phantom nodes, PageRank manipulation, and topological graph vulnerabilities.
March 2009: running code from a stylesheet with XBL in Firefox and HTC in Internet Explorer.
