Luiz Alves · Software engineer
Full stack engineer with .NET, Node.js and Python: systems built around how you work, automations that take repetitive work off your team, and e-commerce platforms made faster, cheaper and more reliable.
Companies that trusted my work
Agência Gentileza
I'm the engineer you bring in when something needs building, automating or fixing, across backend, frontend, infrastructure and AI. These are the problems I see most often. They're examples, not a menu.
Monthly spreadsheets, numbering kept by hand, copy and paste between systems. I turn the process into software built around how you actually work, from requirements to production.
→Receipts system built solo, live on AWS
Reading documents, sorting requests, pulling data out of emails and PDFs, drafting replies. I build automations on top of LLMs where the AI proposes and your rules check it before anything is saved or sent.
→AI entries validated by code before saving
Bots and crawlers can take a large share of what your servers process. I measure your traffic by source and put the right rule in front of each one.
→−49.5% page-rendering load
Idle databases, network endpoints, environments that never sleep. I tie each line of the bill to what it serves and cut what doesn’t pay for itself.
→~US$86/month of fixed AWS cost removed
I automate the path from merge to production: build only what changed, lean images, servers that update themselves, no SSH to ship.
→Manual SSH deploys → ~3.5 min automated
Marketplace, payment and ERP integrations that drop orders or stock updates. Webhooks, queues, jobs and retries you can trust, plus alerts when something breaks.
→Tens of thousands of webhooks and jobs a day
What your platform says was sold versus what reached the bank. Statement import, matching rules and a review queue for whatever doesn’t match.
→Reconciliation backend built from scratch
Old and new running side by side, compared automatically and kept in sync until cutover, so nothing gets lost on the way.
→Finance migration with no data lost
Most problems aren't. If it lives in your code or your infrastructure, send it over. I'll tell you honestly whether I'm the right person for it.
Tell me about itSystems I built end to end, followed by work done as a software engineer at the companies listed under Experience. Clients are kept anonymous and figures about their operations are given in ranges.
Four institutes send a monthly spreadsheet of expenses, and the client issued reimbursement receipts from them, with the numbering tracked in a separate sheet. I built the whole system alone, from requirements to production on AWS: import with line-by-line review, a registry of the people being reimbursed, and receipts generated as PDFs in the background, ready to download one by one or in a batch.
Result~US$86/month of fixed infrastructure cost removed, and cold starts down from ~22 s to ~5 s.
Finance app for households on iOS, Android and web. You type an expense the way you would say it and an LLM turns it into a structured entry, with card statements, installments and recurring items handled for you.
Each customer manages connections, contacts and messages, with instant or scheduled sending and real-time per-recipient status.
On a multi-tenant e-commerce platform with dozens of stores and around a million products, I measured where the rendering work went: about three out of four rendered pages were served to bots, not shoppers, and five SEO crawlers alone accounted for half of it. I blocked those by user agent in nginx across every store, blocked AI and social-media crawlers on the image endpoint (plus an old scraper pulling tens of thousands of images a day), put smaller bots on a crawl delay and took internal search out of the index. I evaluated rate limiting and dropped it: the remaining volume didn’t justify it.
Result−49.5% page-rendering load, with no new infrastructure.
The legacy platform was deployed by hand, over SSH, folder by folder. The new pipeline builds only the service that changed, ships lean standalone images (~195MB instead of the whole monorepo) through Jenkins and Docker Hub, and each server pulls the new version on its own every couple of minutes, recreating only the containers that changed.
Result~3.5 min from merge to production. Nobody logs in over SSH to deploy anymore.
Integrations with large Brazilian marketplaces and e-commerce platforms, with async processing on Redis, RabbitMQ and Go workers. I also traced intermittent runtime crashes in production and moved the affected services to Node.js.
ResultTens of thousands of webhooks and background jobs a day.
At a software house, I built custom systems for clients from zero to production. One was a financial reconciliation backend that imports bank statements, normalizes them, matches entries automatically and sends only the divergences to human review. Others included a help desk with real-time chat and SLA tracking, and payment features (QR code, transfers, virtual cards) added to existing apps.
ResultEach one taken from an empty repository to production.
Replaced a legacy financial system with a new multi-company module in .NET, with tenant isolation through PostgreSQL Row-Level Security. Legacy and new ran side by side with automated comparison and stayed in sync until cutover. Sales now post to finance in the same transaction, with idempotent automatic settlements. I also recovered tens of thousands of products and the settings of dozens of stores from the old platform.
ResultCutover with no data lost.
Send me an email describing the problem. I reply in writing, usually within one business day, with questions or a first read.
Once I understand the problem, you get a short proposal: what I'll do, what you'll get, timeline and a fixed price. Ongoing work can be a monthly block of hours instead. Nothing starts before you approve it.
I'm in Natal, Brazil (UTC−3): 2–3 hours behind the UK and Ireland, and 1–2 hours ahead of US Eastern. Written updates as the work moves; calls when they actually help.
Code, docs and access stay with you. I leave the system documented so your team can run it without me.
Multi-tenant e-commerce platform and an ERP with a financial module in C#/.NET.
Software house. Custom systems for clients: financial reconciliation, help desk, payment features, SaaS frontend.
Frontend for online stores and the company website.
Certification platform from day one to production, with AWS Cognito auth and PDF certificates.
Available for contract and project work. Tell me what's going on and I'll reply in writing.