Company by company
The trajectory in order, with context that a resume cannot hold.
GF Casa Decor
Feb/2018 - Mar/2019In high school I was not very engaged with studying, and for a while I confused that with not liking to learn. I discovered the opposite in my Business Management degree: when the subject interested me, I learned quickly and got genuinely involved. I enjoyed the strategic side of the course, from logistics to customer understanding, and did well in exams, projects, and presentations.
That is where the opportunity appeared. A classmate owned GF Casa Decor and invited me to join the team.
GF was a very small company at the time, three partners, and I was one of the first employees. Because it was small, I did everything: customer service, sales, and administrative routines.
That is where I learned, in practice, how a company works from the inside and the weight of having real responsibility. I left with a foundation in client relationships and communication that, years later, proved just as important as the technical side in Professional Services.
I stayed about a year and a bit. I left when I passed the Nubank selection process. I was a fan of the company and wanted to get in any way I could, for any role.
Nubank
May/2019 - Jul/2021I came in through the door many people used to build their careers there: Customer Experience. I performed well in that area, and at a moment when the fraud team was facing a large fraud wave and urgently needed people, they opened an internal selection. They chose those with the best performance cycle scores for a one-month trial in the anti-fraud area.
Because of the urgency, it was not a typical selection process. It was learning by doing, one day in each area. For me it was practically a promotion. I went, liked the work a lot, and stayed.
I started by investigating fraud cases day to day: KYC, document analysis, transaction monitoring and compliance, at scale. I had strong performance in case resolution, and that was my first serious contact with SQL and indicator analysis.
My first real project was born in that period, Apolo. The outsourced teams doing triage needed to ask us questions and there was no dedicated channel for it. Inspired by an initiative from another area, I built from scratch a simple system: a form that fed into a well-structured spreadsheet, through which we received and organized the questions. Solving that problem was when I discovered I enjoyed working with data and building tools to unblock processes.
I was promoted to Analyst II, started doing quality control on the team's analyses, and handled the most complex fraud cases. I also supported the implementation of a facial recognition product, an AI model running on Nubank's base, validating whether it worked and recognized the right people.
There was a cut in my area and I ended up leaving. It was hard, because I really liked Nubank.
AllowMe
Oct/2022 - Apr/2024Between Nubank and AllowMe there was a gap of about fifteen months, and it was intentional. The fraud analyst roles that came up were, at that point, a step back for me in terms of compensation and type of work, so I preferred to wait for the right opportunity.
I used the time to prepare. I took SQL and Python courses and built, on my own, a data analysis project about Corinthians, the team I support. I brought that project to the AllowMe interview, and it helped demonstrate that I already had a stronger technical base. That is how I got in.
This was my entry into Professional Services. I started implementing Identity & Fraud solutions end to end, side by side with client developers in REST API and SDK integrations, running technical kickoffs and troubleshooting, from commercial handoff to production go-live.
It was a demanding transition, in a good way. I came from fraud analysis and all at once had to learn a lot of new things: reading implementation documentation, using Postman, genuinely understanding what an API and an SDK are, and integrating with product and engineering teams I had never worked with before.
It was also my first contact with B2B clients, after an entire career in B2C. I had to learn to run meetings and communicate with very different profiles, from a developer at a small startup to a project manager at a large enterprise, adjusting tone for each one. The work stopped being sequential and operational and became strategic: I was thinking about the client's full flow, which required much more organization and discipline to keep track of several projects at once.
I was lucky to find very good and very patient people who helped me make that shift.
AllowMe was acquired by Serasa Experian, and I came along. Since my work continued the same, the departure had no drama. It was more of a continuation than a break.
Serasa Experian
Jul/2024 - PresentI arrived at Serasa through the AllowMe acquisition, and this time the transition was smooth because the work continued the same. What changed was the scale. I joined a much larger team and was promoted quickly to senior, a promotion that had already been in motion at AllowMe. There I was the least senior in the group. At Serasa I became one of the technical references, with several mid-level and junior analysts around me.
I am the technical reference for a client portfolio throughout the entire implementation, from technical kickoff to production go-live. I run about 20 concurrent projects and technical meetings in English with teams in the United States, India and China, and investigate bugs together with product and engineering down to the root cause.
A good part of my time is also spent unblocking people, something I was already doing back at Nubank when training triage teams. When a new product arrives, I am usually one of the first to understand it thoroughly and then train whoever asks for help, including from other teams. I support the Sales team by explaining the technical side of the products, increasingly in English, and I contribute a lot of input in product team discussions.
The technical leap was significant. At AllowMe the core product was Device Intelligence, my specialty, and Liveness was secondary. At Serasa I had to learn the entire portfolio while simultaneously teaching the AllowMe products to many people. Add to that the routine of an acquisition, where SDKs and APIs change as the best of both companies gets merged, and adapting to change became part of the job.
It was in that context that I created the Serasa Studio Identity & Fraud, the internal platform that centralizes testing, demonstrations and troubleshooting for the anti-fraud products, recognized by leadership and approved for corporate publication.
How I make things happen
Domain knowledge defines what is worth building. AI defines how fast that becomes reality. Here I explain how both work together in my day to day.
I have a technical base I built over the years: SQL, Python, JavaScript, React. More important than the list of languages is what that base gives me: the ability to understand the problem before I start building, to talk with engineering as a peer, and to know what is feasible before proposing anything.
Domain knowledge in Identity & Fraud defines what is worth building. AI is what makes it possible to build it. When the two combine, the result is a tool that solves a real problem, not a technical exercise.
Serasa Studio is the most complete example. I knew exactly what the problem was: fragmented knowledge, demos that depended on whoever knew Postman, troubleshooting with no centralized environment. I knew how it should work, which products needed to be there, what the flow of each would look like. With that clarity, AI helped me turn that understanding into working code, in React and Node.js, covering eleven products with real API calls.
The same reasoning applies to smaller projects. I have a hobby as a DJ and use Python to automate parts of the process: organizing files, fixing metadata, assembling videos. Smaller problems, same logic: start from a concrete problem, know what the solution needs to do, and use AI to make it happen.
The question that interests me is not which languages I know, but which problems I can solve and how fast. With AI as an accelerator, the bottleneck is no longer technical execution, it has become the quality of understanding the problem. That is exactly where six years in operations, implementation, and direct client contact make a difference.
I want to keep going in this direction: using AI to build increasingly useful tools, inside and outside of work, always starting from a real problem I understand well.
Everything I know
Not a keyword list. A map of what I can actually do, with enough context to understand where each skill came from and how it shows up in the work.
I take over the client after the contract is signed. From there, I understand their scenario before any technical meeting, run the kickoff, stay alongside the developer during integration, and follow through to production go-live. I have done this about fifty times. Today I manage around twenty projects at once, at completely different stages.
What I have learned most in this format is adjusting my tone based on who is in the room. With a developer, I debug alongside them. With a bank project manager, I translate.
Liveness, Facial Biometrics, FaceMatch, Device Intelligence, Document Forensics, Document Capture, OCR, Background Check, Fraud Score, LVS and KYC. I know them in the sense of having integrated, trained, demoed, and investigated bugs in each one, not just read the documentation.
Before implementing anti-fraud solutions, I was operating them. KYC, document analysis, transaction monitoring, team analysis quality control. That was at Nubank, at scale. When a client describes an operational problem, I understand what they are feeling before they finish the sentence. That experience has no substitute.
SQL and Python day to day, for analysis, automations and scripts. REST APIs, SDKs and Postman for integration and troubleshooting. React and Node.js for building. With AI as an execution partner, Serasa Studio is the biggest example of what I can do with that combination.
I run kickoffs, troubleshooting sessions and product alignment in English with teams in the United States, India and China.
Training shows up since Nubank and I carry it forward. When someone from another team needs to understand a product because their clients are about to start using it, that knowledge comes through me. I enjoy it. I learn more by teaching than by being taught. I also have a lot of contact with the Sales team, explaining the technical side of products to clients who are still evaluating.