João Ferreira

Senior Software Engineer

AI, High Performance Computing & Algorithms

Summary

Senior Software Engineer with 5+ years of experience, having started programming C++ at 13 and Reinforcement Learning at 16, specializing in High Performance Computing (HPC), AI/ML, and scalable distributed architectures. Expertise in low-level optimization using C++, CUDA, and Rust to develop efficient, production-ready AI models, designing custom neural network architectures for real-time heuristic evaluation and high-throughput inference engines. Proven track record in real-time systems, such as multi-threaded neural network optimizations and configurable inference frameworks, ensuring predictable performance under high concurrency. Architected high-stakes distributed solutions on AWS EKS and GCP. Passionate about algorithm-level innovations and multi-agent AI.

Professional Experience

Research Grant

University of Aveiro

Nov 2020 - Mar 2022

  • Developed a web platform for medical ML, enabling remote model training, fine-tuning, and analysis.

Software Engineer

Mint State Labs

Mar 2022 - Dec 2022

  • Developed a high-performance Web3 NFT metadata caching service in Rust, designed to serve NFT collection data faster than leading marketplaces at the time.
  • Architected a layered caching stack over MongoDB and deployed the service on AWS EKS to keep latency low and performance stable during traffic spikes.

Cloud & AI Software Engineer

Orange Comet, Inc.

Dec 2022 - Jul 2024

  • Designed and developed microservices in Rust + Actix Web on AWS EKS for public web3 related services.
  • Owned a high-budget GPU rendering pipeline on Google Cloud using Blender CLI to create NFT video collections, planning and monitoring multi-day rendering campaigns.
  • Designed an AI-driven random map generator in C++, later ported to Lua, producing high-quality maps with fair starting conditions for multiple players.
  • Developed and optimized AI behavior modules for Unreal Engine so complex agent logic could run efficiently in real time.

Senior Real Time & AI Engineer

Cint

Aug 2024 - Present

  • Built and optimized real-time systems that power core products under high concurrency.
  • Contributed to strategic decisions on adopting and integrating AI services across the platform.
  • Planned, developed, and deployed multi-threaded optimizations for low-level neural-network instruction models on shared real-time environments.
  • Created runtime-compilable, parallelization-aware inference frameworks for configurable and highly efficient low-level instructions.
  • Developed knowledge-retrieval services and internal MCP servers to expose company knowledge to AI assistants and internal tools.

Technical Skills

Programming Languages

C++CUDARustPythonTypeScriptJava

Artificial Intelligence

Deep Reinforcement LearningMCTSLLM Deployment & Fine-TuningCompressed Inference ModelsCUDA Model TrainingInstruction Model Frameworks

Systems & Performance

Algorithm OptimizationMulti-ThreadingGPU AccelerationDistributed Systems

Cloud & Infrastructure

AWSGCPKubernetesCI/CDReplica Async Synchronization

Personal Successful Projects

Here are some of my key personal projects. These examples demonstrate my ability to apply advanced algorithms and build high-performance systems from the ground up to solve complex challenges in competitive programming, AI, and automated investments.

Custom Deep Learning Frameworks

Authored lightweight, TensorFlow-like frameworks in Rust, C++, and CUDA. These projects were built for high efficient computing production-oriented inference.

High-Performance Board Game AIs

Engineered C++ solvers using Monte Carlo Tree Search (MCTS), achieving 2nd worldwide rank on Connect 4 and a top 0.5% global rank on Ultimate Tic Tac Toe. Solutions integrated bitmaps, compact neural networks, and transposition tables.

Crypto Trading Bot (CUDA/C++)

Trained and deployed predictive models from scratch in CUDA/C++ without external ML libraries to ensure maximum performance and control. Deployed in a live environment with a monitoring dashboard.

University AI Project Award

Awarded 1st place for developing an innovative multi-objective, double-agent tree search system requiring reduced bit level communication and advanced path finding techniques.

Education & Certifications

My academic background and professional certifications provided a strong foundation in computer science and intelligent systems. It allowed me to explore broad areas related to efficient computation.

Education

Post-Graduate, AI & Robotics

University of Aveiro, 2022

Master's Degree, Computer Science

University of Aveiro, 2015 - 2020

CV updated at 17/11/2025