Ahmad S.

Deep Reinforcement Learning, Robotics, VLA, Humanoid

With 7+ years of experience in Artificial Intelligence, I design and deploy Generative AI, Agentic AI, Deep Reinforcement Learning, and Large Language Model–driven autonomous solutions for robotics, business automation, decision intelligence, and algorithmic trading. My work spans Deep Reinforcement Learning, Foundation Models, autonomous robotics agents, and robotics decision frameworks. I specialize in edge AI deployment using ROS1/ROS2, Jetson Nano/Xavier/Orin, and Raspberry Pi, delivering end-to-end pipelines from simulation to real-world hardware. Simulation expertise includes PyBullet, Gazebo-ROS, Unity3D, and PyGame, supporting 2D/3D environments for aerial, ground, and marine robotics with integrated AI model deployment on Pixhawk, Jetson, and Raspberry Pi platforms. My core strengths include ML/DL/CV model development, fine-tuning, customization, POC prototyping, and scalable AI system design. I am also open to research collaborations in Deep Reinforcement Learning and embodied AI. ## Agentic AI, AI Agents & RAG Tooling LangChain AI agent orchestration, tool-using AI agents, and multi-step reasoning pipelines LlamaIndex RAG-based AI agents, knowledge-aware AI systems, and retrieval-driven AI workflows AutoGen collaborative AI agents, conversational AI agents, and multi-agent task execution OpenAI function-calling / Assistants APIs – production AI agents with tool integration and decision-support AI agents Supporting AI Agent Infrastructure Vector Retrieval for AI Agents: FAISS, Chroma for memory-augmented AI agents and RAG-based AI decision systems Deployment for AI Agents: FastAPI and Docker for scalable AI agent services and autonomous AI pipelines LLM Integration: prompt engineering, fine-tuning, and evaluation of LLM-powered AI agents and knowledge-generating AI systems ## Deep Reinforcement Learning Expertise * DQN, Double DQN, Dueling DQN * PPO, SAC, DDPG, TD3 & custom RL architectures * Model-based and model-free RL * Federated & distributed RL * Multi-agent, competitive, and collaborative RL * Swarm intelligence & data-driven RL * RL integrated with LLM reasoning and planning * Custom OpenAI Gym/Gymnasium MDP design * Simulation-to-real transfer and hardware deployment ## Agentic AI & Autonomous Robotics Decision Intelligence * Agentic AI architectures for robotics and enterprise automation * Robot policy learning and adaptive control * LLM-driven task planning and hierarchical reasoning * Language-image goal-conditioned learning * Robot Transformers & embodied foundation models * In-context decision learning and tool-using agents * Open-vocabulary navigation and manipulation ## Robotics Perception & Embodied Intelligence * Open-vocabulary object detection, 3D classification & semantic segmentation * Person following, tracking, and identification * Visual navigation, visual SLAM, and multimodal perception ## Generative AI, RAG & Knowledge Systems * Retrieval-Augmented Generation (RAG) pipelines * Knowledge graph construction and knowledge generation workflows * LLM fine-tuning, evaluation, and alignment * Autonomous AI copilots and decision-support agents * Multimodal generative models and AI orchestration APIs --- ## Frameworks & Tooling RL & AI: Ray RLlib, TorchRL, Stable-Baselines3, MuJoCo, Gymnasium, Unity ML-Agents Robotics & Simulation: ROS1, ROS2, PyBullet, Gazebo, Unity3D, PyGame2D ## Hardware Platforms Jetson Nano, Xavier, Orin • Raspberry Pi • Arduino • STM32 ## Robotics Platforms Experience Unitree GO1 (Quadruped), Hiwonder JetHexa (Hexapod), Ackermann & Differential Drive Robots, Manipulators, Soft Robotics, Micro-robots I welcome collaboration opportunities in autonomous robotics, agentic AI, reinforcement learning, and generative knowledge systems. Let’s discuss how my technical depth can accelerate your project from concept to deployment. Regards Engineer Suleman Autonomous AI & Decision Support