Leandro Lacerda
Machine Learning Research Scientist at Inter
Minas Gerais, Brazil
Summary
👤
Senior🎓
Top SchoolLeandro Lacerda is a Machine Learning Research Scientist with eight years of experience building probabilistic deep learning systems and production ML pipelines. Based in Minas Gerais, Brazil, he blends rigorous academic training (MSc and ongoing PhD in Computer Science) with hands-on engineering across JAX, PyTorch, TensorFlow and distributed AWS stacks to deploy models on GPU/TPU at scale. At Inter he drives research on recommendation systems, time-series forecasting, AML detection and latent-variable inference while automating MLOps with SageMaker Pipelines, Lambda and Terraform. As an open-source contributor, he implemented the Skew Normal distribution in TensorFlow Probability, adding numerically stable CDF/quantile logic and derivatives useful for probabilistic modeling. Earlier roles span analytics-driven product impact—cutting churn ~45% as a data science lead—and public innovation, where he co-created a major startup accelerator in Latin America. He combines a mathematician’s precision with product-minded execution, often bringing production-grade research prototypes into real-world use.
8 years of coding experience
10 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at Universidade Federal de Minas Gerais
English, Portuguese
Github Skills (7)
statistics10
probabilistic-programming10
tensorflow10
python10
machine-learning9
bayesian-methods8
data-science8
Programming languages (6)
Github contributions (5)
tensorflow/probability
Feb 2022 - Dec 2022
Probabilistic reasoning and statistical analysis in TensorFlow
Role in this project:
Data Scientist
Data ScientistContributions:24 reviews, 67 commits, 6 PRs in 9 months
Contributions summary:Leandro contributed to the implementation of the Skew Normal distribution within the TensorFlow Probability library. Their work involved defining the mathematical properties, including the PDF, CDF, and quantile functions, as well as creating the necessary functions for sampling and calculating the log probability. The contributions also included the addition of partial derivatives for the inverse cumulative distribution function (CDF) calculations, and unit tests to ensure the implementation's accuracy and numerical stability.
statisticspythonprobabilistic-reasoningdata-sciencedeep-learning
leandrolcampos/coursera-dsa
Jun 2018 - Apr 2019
Some solutions for Data Structures and Algorithms Specialization at Coursera.
Contributions:75 commits, 1 push in 10 months
algorithms-specializationbinary-search-treecppspecializationalgorithms-and-data-structures
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.