Skip to content
View gabriel-lab-ia's full-sized avatar
  • UniCEUB University
  • Brasília🇧🇷

Block or report gabriel-lab-ia

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
gabriel-lab-ia/README.md

Gabriel Duarte. Machine Learning / Reinforcement Learning Engineer (Early-Career) • GPU Systems • Robotics Simulation • Autonomous Systems Based on Artificial Intelligence

BSc in Artificial Intelligence (1st Semester). Engineering trajectory focused on Reinforcement Learning, GPU-accelerated physics simulation, and low-latency inference pipelines built on the NVIDIA ecosystem.

Work centered on control, optimization, and systems engineering, integrating TensorRT, CUDA, Isaac-based physics engines, and modern robotics frameworks.

Essential technical areas under daily development with peak performance: Reinforcement Learning Engineering PPO / SAC / TD3 Policy export: PyTorch → ONNX → TensorRT FP16 Latency-oriented benchmarks (CPU vs CUDA vs TensorRT) Real-time continuous-control evaluation GPU & High-Performance Computing CUDA Toolkit 12.0 + CUDA Runtime 13 PyTorch CUDA TensorRT 10 (builder, engines, FP16 optimization) Parallel simulation and vectorized training workflows Simulation & Robotics NVIDIA Isaac Gym MuJoCo 3.5 Control-loop engineering ROS2 (Jazzy) EdgeAI & Deployment Low-latency inference design ONNX Runtime CUDA Real-time RL system execution Model compression and runtime optimization

Academic Formation: BSc in Artificial Intelligence — UniCEUB (Campus Asa Norte) Current semester: 1st semester Formal academic curriculum providing foundational competencies in mathematics, computation, algorithms, software engineering, data systems, cybersecurity, cloud, machine learning, and artificial intelligence.

This degree provides the theoretical and computational base for AI systems, while my Independent Engineering Track (below) covers advanced Reinforcement Learning, GPU systems, simulation, robotics, optimization, and deployment engineering beyond the curriculum.

Academic Curriculum Structured:

• Semester 1 ~ Conversational Agents ~ Linear Algebra and Analytic Geometry ~ Bootcamp I ~ Software Engineering ~ Programming Logic ~ Probability and Statistics

• Semester 2 ~ Calculus for Computing ~ Cloud Computing ~ Ethical Dialogues and Reflections ~ Cybersecurity Fundamentals ~ Foundations of Artificial Intelligence

Semester 3 ~ Machine Learning ~ Database Systems ~ NoSQL Systems ~ Bootcamp II ~ Advanced Calculus for Computing ~ AI Development I ~ Digital Signal Processing for AI

Semester 4 ~ Systems Analysis and Design ~ Deep Learning ~ AI Development II ~ AI Data Structures ~ Project Management ~ Artificial Neural Networks

Semester 5 ~ Algorithm Analysis and Design ~ Bootcamp III ~ AI Entrepreneurship ~ Natural Language Processing ~ Integrative Project I ~ Computer Vision

Semester 6 ~ Unsupervised Learning ~ High-Performance Computing (HPC) ~ MLOps ~ Elective I < Integrative Project II

Semester 7 ~ Bootcamp IV ~ Generative Artificial Intelligence ~ Elective II ~ Integrative Project III ~ Special Topics in Artificial Intelligence

This academic track establishes the formal baseline: Mathematics for AI (algebra, calculus, probability, optimization) Software engineering and computational thinking Data systems and distributed computation Machine Learning fundamentals Neural networks and deep learning Cybersecurity and cloud systems High-performance computing and MLOps NLP and Computer Vision Capstone projects and applied AI design

• Combined with my independent engineering development (Reinforcement Learning, Isaac Gym, CUDA 13, TensorRT, MuJoCo, ROS2, low-latency control, GPU systems), this forms a dual-track training approach: one academic, one technically aggressive and industry-aligned.

Technical roadmap for individual performance during the university period: Mathematics for RL (linear algebra, optimization, probability) CUDA runtime behavior, profiling, and memory models TensorRT optimization workflows Isaac Gym fundamentals ROS2 operational basics Advanced continuous control Massive GPU-parallel simulation (Isaac Gym) Distributed training pipelines Navigation/manipulation in ROS2 environments Custom CUDA kernels Specialized physics simulators EdgeAI deployment (Jetson Orin) TensorRT + Triton end-to-end pipelines Large-scale RL infrastructure Low-latency robotic control systems Full-stack kinematic/deployment architectures End-to-end optimization of models + simulators

Pinned Loading

  1. rl-mujoco-trt-ppo-hopper-low-latency rl-mujoco-trt-ppo-hopper-low-latency Public

    Low-latency PPO reinforcement learning pipeline for MuJoCo Hopper, optimized end-to-end with CUDA, ONNX export, and TensorRT inference engine. Includes reproducible training setup, benchmarking too…

    Python 1