Eliseu Venites Filho

Computational Statistical Physics Data Science Systems Engineering

Skills

Languages: Portuguese Native English Fluent French Advanced
Programming: C C++ C++17 boost Rust tokio rayon faer Python Pandas NumPy SciPy scikit-learn matplotlib PyTorch TensorFlow Julia DataFrames.jl Plots.jl Makie.jl Haskell megaparsec Delphi
Tools: Databases SQLite PostgreSQL DuckDB Docker Nix

Tools

Databases SQLite, PostgreSQL, DuckDB

Docker

Nix

Experience

Software Engineer

Nelogica Porto Alegre, Brazil
  • Engineer low-latency, high-performance backend microservices in Delphi as part of the Automation Tools team, serving thousands of daily active users
  • Sustain high availability on mission-critical legacy systems through proactive maintenance and infrastructure hardening
  • Reduce manual processing time by automating workflows and building custom internal tools that scale operations across teams
  • Deploy and operate services across AWS, Azure, and Equinix infrastructure, using GitLab CI pipelines within an Agile/Scrum team workflow

Ph.D. in Computational Statistical Physics

Universidade Federal do Rio Grande do Sul Porto Alegre, Brazil
  • Doctoral research funded by a CNPq scholarship
  • Analysis of the ensemble correlations of observables of complex systems in order to predict their critical behavior
  • Systems from different Universality Classes considered
  • Both systems with and without a defined Hamiltonian considered
  • The simulations were implemented in Rust while the data analysis was done in the Julia ecosystem

M.Sc. in Computational Statistical Physics

Universidade Federal do Rio Grande do Sul Porto Alegre, Brazil
  • Master's research funded by a CNPq scholarship
  • Scored higher than 99.42% of candidates on the EUF 2-2018 (National graduate programs entrance exam)
  • Performance evaluation of the Simulated Annealing applied to different configurations of the Traveling Salesman Problem
  • Analysis of the stochastic optimization algorithm applied to problems at the boundary between P and NP complexity classes
  • The optimization algorithm was implemented in C++ while the data analysis was done in the Python ecosystem

Optical Engineering Internship

Télécom ParisTech Paris, France
  • Improved photon-pair coincidence count rate by engineering active stabilization of a polarization-entangled photon-pair source
  • Enabled reliable Quantum Key Distribution (QKD) protocol testing by optimizing signal count rates within the Information Quantique et Applications research group

Undergraduate Research (Scientific Initiation)

Universidade Federal do Rio Grande do Sul Porto Alegre, Brazil
  • Conducted CAPES-funded undergraduate research in Quantum Information
  • Gave presentations on quantum algorithms such as Shor's and Grover's

Education

Diplôme d'Ingénieur

Institut d'Optique Graduate School Palaiseau, France
  • Double degree in the context of BRAFITEC program
  • Optical Instrumentation, Automation, Lasers and Quantum Optics

L3 et M1 en Physique Fondamentale

Université Paris-Saclay Orsay, France
  • Double degree in Theoretical Physics offered to engineering students
  • Analytical Mechanics, Statistical Physics, Plasma Physics and Atomic and Molecular Physics

B.Sc. in Engineering Physics

Universidade Federal do Rio Grande do Sul Porto Alegre, Brazil
  • Graduated Summa Cum Laude with a final GPA of 9.54/10.0
  • Solid State Physics, Electronics, Microprocessors, Photonics and Material Analysis

Projects

tsp-sa C++ | Python

  • Built a modular C++ optimization library implementing Simulated Annealing for the M.Sc. research
  • Supports Generalized Simulated Annealing and Tsallis Entropy statistics
  • Decoupled the optimization core from the TSP implementation, enabling reuse across arbitrary Markov chain problems
  • Automated data analysis and plotting in Python to accelerate research iteration

artificial-systems Rust (ndarray, serde)

  • Implemented high-performance Rust simulations of computational models for the Ph.D. research
  • Modeled Spin Systems (Ising and Blume-Capel models) at scale
  • Investigated the Contact Process with diffusion

ts-cov-matrix Julia (DataFrames.jl, Makie.jl)

  • Applied Random Matrix Theory to analyze time-series covariance matrices for the Ph.D. research
  • Compared spectral properties against the Marchenko-Pastur distribution across NOAA temperature records, Spin Systems, and Contact Processes
  • Built a full data analysis pipeline in the Julia ecosystem (DataFrames.jl, Makie.jl)

json-parser Haskell

  • Strict JSON parser implemented in Haskell using Megaparsec
  • Adheres closely to JSON standards
  • Can be used as a library or a standalone command-line tool

sternhalma-server Rust (tokio)

  • Asynchronous game server for Sternhalma (Chinese Checkers) built with Rust and Tokio
  • Actor-like architecture with decoupled game logic and connection handling
  • Client-agnostic design supporting CLI, GUI, and AI agents
  • Supports both Raw TCP and WebSocket connections using a CBOR-based protocol

sternhalma-agent (WIP) Python (PyTorch)

  • Reinforcement learning agent implementing AlphaZero from scratch
  • Uses Monte Carlo Tree Search (MCTS) for planning and Deep Neural Networks (ResNet) for evaluation
  • Designed to master Sternhalma through self-play without human knowledge

Publications

Efficient computational method using random matrices describing critical thermodynamics

R. da Silva, E. Venites Filho, S. D. Prado, J. R. Drugowich de Felício
International Journal of Modern Physics C

Mean-Field Criticality Explained by Random Matrices Theory

R. da Silva, H. C. M. Fernandes, E. Venites Filho, S. D. Prado, J. R. Drugowich de Felício
Brazilian Journal of Physics