Start with the expensive problem.
Missed orders, wrong payroll answers, weak access controls, repeated UI bugs. The cost of doing nothing tells me whether the work matters.
// the short version
I take expensive software problems all the way to production.
Product, security, infrastructure, and AI systems where a confident mistake costs more than a slow answer. I build the system, lead the team, and stay for production.
Missed orders, wrong payroll answers, weak access controls, repeated UI bugs. The cost of doing nothing tells me whether the work matters.
A product should make sense in one screen or one short flow. If the result needs a long explanation, the product still needs work.
Independent work runs side by side. Real dependencies stay ordered. Shared files, approvals, and merge points get explicit owners.
Tests that ran, citations a person can open, audit records, production logs, and human review. A model agreeing with itself is not proof.
Free tiers are fine for experiments. Customer data, uptime, privacy, and predictable capacity need a production decision and a fallback.
One shipped system can become a case study, a field note, a demo, a hiring story, and a sales answer. The work stays the source.
This is the short version. The work is public; the personal details are not. The black bars keep it that way. Tap or hover them.
I am the CTO at Nourishubs, the founder of Bienzoli, and a senior engineer at SD Worx. I build products, lead technical teams, and deal with production when it gets messy.
Chief Technology Officer (Contract)
Founder
Senior Software Engineer
Full Stack Software Engineer
Junior Front End Developer
Web Developer
Web Developer
// the problem
Nourishubs handles children's data in 4 countries. Each country has its own privacy law, so one generic compliance checklist was never going to be enough.
// what he did
// why it matters
I led the security work at 22 while running the 10 person engineering team. Reading the code made the policies more realistic, and writing the policies exposed gaps in the code.
proof
Risk register, access records, control owners, and audit preparation across four countries.
constraint
One management system had to respect four different national privacy laws.
// the problem
HR and payroll answers sit across dense policy documents. Staff lose time finding the right rule, and a confident wrong answer can affect someone's pay.
client internals: ████ nda ████
// what he did
// why it matters
This is still in development inside a major HR and payroll company. The work is mostly about retrieval quality, clear refusals, and repeatable testing. The chat interface is the easy part.
proof
A fixed question set measures confident wrong answers. Every supported answer cites the policy source.
constraint
In progress and under NDA. No client internals or invented performance claims.
// the problem
Teams were rebuilding the same interface patterns across projects. That created design drift, repeated bugs, and avoidable review work.
// what he did
// why it matters
Teams can build new screens from known parts, and fixes apply across the products that use them. The library also gives junior engineers a clear standard to follow.
proof
The shared tokens, primitives, and patterns are used across products and by the 3 junior engineers I mentor.
constraint
The record proves adoption, not an invented percentage for time saved or defects removed.
// live in production · click through
// personal builds
Platform that closes the gap between theory and shipping. Real projects, real mentorship, front to back.
Sports betting research and bankroll tracker that collects trusted tips and curates daily picks.
Job board with tailored matches, employer profiles, and detailed filtering.
Social platform for conservation. Auth, donations, cloud media. IEEE Hackathon winner.
Travel booking built on Amadeus APIs, with live flight and hotel data plus secure payments.
Pet adoption and donation platform with campaigns, a vet finder, and assisted pet matching.
Ask about my work, projects, or experience. Answers come from a fixed dossier. Direct questions get direct answers.
// awaiting query
generated answers use a fixed dossier. anything else is redacted.
Each note starts with work I have done: a system shipped, a failure found, or a decision I had to defend. The same evidence becomes a case study, a short explanation, or a direct answer when someone asks.
Paste a job description and compare it with my actual record. The result gives you a match score, supporting evidence, and the gaps I would need to close.
the machine is waiting.
Send the problem, who it affects, and what failure costs. I will answer directly. I ignore vague solicitation.