Resume

Professional Experience

Vulkan Labs

Founder July 2024
Building the decision infrastructure to power the next decade in Finance.

Nubank

Machile Learning Engineer August 2023 $\rightarrow$ July 2024

C6 Bank

Machine Learning Engineer May 2022 $\rightarrow$ July 2023
1. Developed and implemented services to minimize model deployment time and knowledge barriers, enabling data scientists to focus on ML tasks. Reduced average deployment time from 1-2 weeks to 2 days.
2. One of two developers of the model inference infrastructure and architecture, serving millions of predictions a month with 1-3s latency and auditable data trails. Used Go serverless functions and Vertex AI models and maintained a stable infrastructure since 2020, from the 1st to the 20th-million customers, from 1 model to 5+ versions a quarter.
3. Managed the analytics development platform, a service for all of the Bank's users, which allows users to create dev servers on demand, currently serving 100+ weekly users across multiple teams. Utilized GKE and Terraform to provide a less restrictive environment that is still secure and compliant.
4. Implemented multiple data services:
  - Low-latency, cost-effective feature data retrieval, serving ~50M data points per month.
  - Geolocation-based attribute querying using a parallelized RTree to serve ~65M segments and 10 attributes with <200 ms latency.
  - Data ingestion for the Analytics team, using Apache Beam & Dataflow to process multiple terabytes of data for the bank's Data Lake.
Data Scientist Jan 2019 $\rightarrow$ April 2022
1. Application and Transactional Fraud
  1. Developed first fraud prevention policies, identifying vulnerabilities in the process and automating most of the account opening risk assessment. Proposed of many of the current features, such as liveness detection.
  2. Created models to evaluate the risk that an account would be used to facilitate transaction fraud or money occultation.
  3. Helped evaluate and improve transactional fraud prevention policies, rewriting much of the policies.
2. Consignado (Payroll Loans)
  1. Implemented policy evaluations and many of the business indicators that were addopted for the product, including early fraud indicators and definitions of salesperson efficiency and performance.
  2. Developed first models for the product, working with individual proposals and salesperson behavior.
3. Tooling
  1. Started internal library for standardized, declarative analyses, improving communication standards and allowing data scientist to share a codebase for common cases
  2. Developed standards on model development, deployment, architecture that led to previously mentioned work
  3. Collaborated on shared libraries for model evaluation and quality monitoring
4. Onboarding
  1. Analyzed and streamlined process, eventually cutting much of the manual work in favor of automated verifications.
  2. Developed first policies for identity verification and information validation that allowed automated approval of most proposals.
5. Chatbot
  1. Implemented interaction with bank services, allowing users to solve complex issues instead of just asking questions.

Daitan Group

Data Scientist Jul 2018 $\rightarrow$ Jan 2019
1. Unsupervised behavioral models for detecting and preventing financial fraud. Participated in architecture modernization and model updates, broadening the range of frauds caught and providing more flexible modeling tools
Data Science Intern Feb 2018 $\rightarrow$ Jun 2018
1. Participated in a team of data scientists developing solutions for detecting anomalies in telecommunications data
Software Engineering Intern Oct 2017 $\rightarrow$ Jan 2018
Software Testing Intern Mar 2017 $\rightarrow$ Oct 2017

Academic Background

Unicamp - University of Campinas

FEEC - School of Electric and Computer Engineering

Graduate classes - Aug 2017 $\rightarrow$ Jun 2019
- Convolutional Neural Networks for Image Analysis
  Final Project: Explored the differences between various architectures and training regimes by training a simple GAN (DCGAN with Wasserstein distance).
- Information Theory
- Stochastic Processes
- Adaptive Signal Processing
Bachelor of Electrical Engineering Feb 2013 $\rightarrow$ Jul 2018
Graduation Project: Implementing Capsule Networks in PyTorch

#about