Hi, I'm Lucas Rosa
Backend Developer · Node.js · TypeScript · PostgreSQL
APIs, integrations and data syncs in production, with tests and monitoring. I built the platform that tracks each partner from the first click to the revenue they generate. I come from Growth and data, so I start with the problem.
How the platform works
from ad to revenue, in one place
Comes in from a campaign, already tagged with its origin.
96.8% of signups linked to their original lead, no spreadsheets.
Production tracked month by month: banks, products and ticket size.
Marketing sees how much each channel generated, not just lead counts.
96.8%
of partners linked to origin
automatic RD Station × CRM match
~US$ 400
saved per year
social reports without the Reportei subscription
22
competitors tracked
every week, for ~US$ 3/month
7
integrated systems
CRM, RD Station, Meta, AI and more
Featured case
01 / 03
01 · Problem
Marketing stopped at the lead: it didn't know which partners came from campaigns or what they produced afterwards. And it paid for a tool just to get social media reports.
02 · What I did
I built the platform end to end: integrated RD Station and the CRM through their APIs, tracked each partner's production and created a social media module powered by AI.
03 · Result
Marketing now knows which partners it brought in and how much they produced. The Reportei subscription was cancelled, and AI now analyzes every post.
Inside the platform
From signup to results: how much each partner produces and where they came from.
Replaced the Reportei subscription and cut the team's costs.
Competitors tracked every week for ~US$ 3/month.
The marketing funnel measured end to end.
The foundation: partners and contracts synced automatically, no spreadsheets.
Custom login with an audit trail.
Architecture
Data comes in through scheduled functions, everything goes through Postgres, and the dashboard only reads through the backend.
External sources
Cron jobs run every 2–3 minutes in chunks of up to 120 s and save their progress. If something fails, the next run resumes where it stopped. Health is checked against the stored data, not the cron status.
Internal tables use deny-all RLS: the dashboard reads only through server functions with its own sessions. Secrets live in the database Vault, and every cron authenticates with its own key.
A monitor checks the syncs every 5 minutes, and a daily contract audit posts a severity report to Discord before anyone notices a wrong number on the dashboard.
Live sites
Drag, use the arrows or let it play.
Product landing page
Payroll loan product, hand-coded, with simulation and WhatsApp checkout.
Service landing page
Post-surgery physiotherapy: services, testimonials and booking.
Conversion landing page
FGTS advance product with Meta Pixel and Tag Manager.
Corporate website
Company website: services, mission and testimonials.
Projects
Stack
About
Istartwiththebusinessproblemandonlystopwhenthesystemisinproduction,solvingthatproblem.
I worked in Growth, Inside Sales and BI (CAC, LTV, funnels and Python ETL) while taking on freelance development projects in parallel.
I started programming to integrate systems and automate processes. I use AI as a tool and make a point of understanding every technical decision.
Back-end Developer program, SENAI/SC · graduating Dec 2026
296 contributions in the last year
66 days with commits · steady pace since March 2026
Recommendation
“He isn't satisfied with just making it work: he wants to understand why and how it worked. When we reviewed code together, he took the feedback, applied it the next day and came back with even better questions.”
