Development · Research · Community

Ponta Grossa, Brazil

Victor
Cerqueira.

/ Bnk
Portrait of Victor Cerqueira
Between code and people.

I am a mid-level full-stack developer and Computer Engineering student at UEPG. I work on business systems, APIs and infrastructure. Through research and personal projects, I explore how to build more reliable software — and share what I learn with the community.

01 Projects

Ideas that become systems.

Four parts of my work: a writing tool, a client, a team project and an exploration of development with agents.

01Personal project · Academic writing

ZLEditor

LaTeX, references and PDF previews in one place. A local environment for writing and reviewing academic work.

Python / FastAPI / CodeMirror / LaTeX

Inside the projectZLEditor
The context
Writing a paper means switching between an editor, compiler, bibliography and PDF. I wanted to bring this workflow closer to the experience of a code editor.
My contribution
I developed the browser-based editor, local compilation, PDF preview and bibliography workflows, with optional Zotero integration and AI assistance.
Technical decisions
Python/FastAPI and CodeMirror 6 separate the local service from editing. LaTeX runs on your own machine; AI assistance is optional and depends on the configured provider.
Where it stands
In development since April 2026, with public code. Editing, compilation, previews and bibliography integration are already part of the tool.
Screenshot of the public Cergat Group website

02Client work · Web and systems

Cergat Group

From an institutional presence to internal processes: a live website, signature automation and an evolving CRM.

Next.js / Java / Spring / React / Docker

Inside the projectCergat Group
The context
Cergat needs to present its services and organize administrative and sales workflows. I have worked with the company as a client since June 2023.
My contribution
I developed the institutional website and its infrastructure, built a signature digitization system and am working on the CRM for customers, requests and sales proposals.
Technical decisions
Next.js and React for the website. For the CRM, Java/Spring and React, with domain modules, access control, change history and protection against concurrent edits.
Where it stands
The website is live. The CRM already implements authentication and customer, request and proposal workflows; it continues to evolve with business needs. The internal systems have private code.

03Team project · Observability

LOGAAS

A path from receiving a log to investigating a request, with data isolation between organizations.

Go / RabbitMQ / NestJS / MongoDB

Inside the projectLOGAAS
The context
Distributed services produce records in many places. LOGAAS brings log and trace ingestion, processing and querying together in a multi-tenant platform.
My contribution
I contribute to the team’s development work, focusing on the backend and the ingestion and processing pipeline. The product is a collective effort and its code is private.
Technical decisions
Go for ingestion, RabbitMQ between services and a NestJS API. Publication acknowledgments, reprocessing and tenant isolation guide failure handling.
Where it stands
The backend for ingestion, persistence, querying and authentication is implemented. A foundation of interface components exists; screens, alerts and some interpretation features remain planned.

04Personal tool · Software engineering

SMSDD

Specifications, context and evidence to maintain continuity when developing with agents.

Rust / CLI

Inside the projectSMSDD
The context
When developing with agents, decisions and acceptance criteria can get lost between sessions. SMSDD organizes that context alongside the work.
My contribution
I am developing a Rust tool to record changes, coordinate stages, maintain implementation contracts and collect validation evidence.
Technical decisions
Local artifacts and explicit criteria connect intent, implementation and verification. The workflow makes decisions and results available for inspection across sessions.
Where it stands
An evolving tool described in a public blog post. The post presents the approach and experience of using it; it is not a peer-reviewed academic publication.

02 Experience

Software in production. Learning along the way.

Financial systems, business operations and the responsibility of maintaining what already works.

Mar 2025 — present

MM22 / Grupo Schaffen

Full-Stack Developer

Junior → Mid-level in February 2026

At MM22 / Grupo Schaffen, I work on development, service architecture and infrastructure for the financial and Agrotech sectors.

I provided technical leadership for a financial reconciliation platform that reduced bank audits from around three months to a few hours, using data pipelines and discrepancy analysis in Python.

I developed a B2B Agrotech marketplace to connect producers and suppliers, reducing the time to obtain quotes for agricultural inputs from months to days. I integrated CAR, Sintegra and Receita for document validation.

Jun 2023 — present

Cergat Group

Systems Analyst · Client

Institutional website and internal systems

I work with Cergat Group on website development, infrastructure and administrative process automation.

Alongside the evolving CRM, I implemented signature digitization with NestJS, RabbitMQ and Python services, reducing manual steps.

One ERP, around 110 clients.

I contributed to the architecture of an ERP serving around 110 clients. The challenge was to accommodate the system’s evolution and different clients’ operations within the same structure, with attention to maintenance and service continuity.

Updating also means knowing how to recover.

I worked on an automated update workflow with recovery in case of failure. The priority was to plan a way back and keep operations running when an update did not complete as expected.

03 Research

Build. Investigate. Share.

University and community are places where I test ideas, ask questions and share what I learn.

Computer Engineering · UEPG · 2023 — present

Current research · since Sep 2025

LIMLA · Compression and machine learning

I investigate how lossy compression of images, audio and video affects ML models. I organize experiments, metrics and adversarial attacks to compare accuracy and robustness in the compressed and original domains.

Undergraduate research · 2024 — 2025

SNPL1DB · Neurodevelopment data

I helped improve an existing database of L1 genetic variants: a Next.js interface, data presentation and new analyses, in collaboration with the research team.

Undergraduate research · 2023 — 2024

SLEKB · Lupus knowledge

I developed web features and helped curate biomedical information for a knowledge base on systemic lupus erythematosus, working with the academic team.

Academic experiment · 2025

BestLanguage

I implemented the Node.js/NestJS API and worked on an experiment with equivalent APIs in containers. I used ANOVA, regression and Tukey tests in R; the results describe the tested conditions, not a universal ranking of languages.

Repository

Implementation experiment

perceptron-mlp

An MLP for solving XOR, implemented in Rust as an exercise to understand how a neural network works.

Repository

04 Writing

Notes along the way.

Decisions, experiments and the work behind the scenes. Articles in Portuguese, on my blog.

Visit the blog