This role within Deep Learning Focus Group is strongly technical, responsible for building DL/AI based solutions for validation of NVIDIA GPUs. For e.g. GPU Render Output Analysis (Video, Images, Audio) and complex problems like Intelligent Game Play Automation. This person would need to analyze/understand the challenges from stakeholders of various groups, design & implement DL solutions to resolve them.
We are specifically looking for expertise in the following key areas:
Game play automation using deep learning (DL) and artificial intelligence (AI) techniques, with a focus on research and application in the gaming industry.The candidate should be proficient in the following DL techniques:
- Deep Reinforcement Learning (DRL) and Imitation Learning
- Vision Transformers (ViTs) and Convolutional Neural Networks (CNNs)
- Generative AI and Diffusion Models
Prior experience or research in AI-driven bots, In-Game Movement Automation, and Player Behavior Prediction are highly desirable.
Knowledge in using AI development tools for test plans creation, test cases development and test cases automation.
What you'll be doing:
- Build Intelligent Gameplay Automation & Agentic Workflows: Design and deploy advanced gameplay agents using computer vision, reinforcement learning, imitation learning, and LLM/VLM-based agents, leveraging state-of-the-art tools like Codex and Claude.
- Drive ML-Driven QA & Defect Detection: Apply ML/DL techniques to solve complex QA challenges across NVIDIA product lines, implementing DL-based solutions for video/audio defect detection and optimizing automated test frameworks to boost productivity.
- Develop End-to-End GPU Validation Solutions: Create and maintain robust, Python-based automation pipelines that consume neural networks to rigorously validate NVIDIA GPUs.
- Establish Scalable Infrastructure & Deployment: Set up and manage scalable development environments using Linux, Docker, and TensorRT to train, validate, and deploy large-scale neural networks.
- Curate Self-Improving Data Pipelines: Build, clean, and augment high-quality datasets to feed and enable continuous, self-improving training pipelines.
What we need to see:
- Master’s or PhD in AI/ML/CS (or equivalent) with at least 2 years of hands‑on ML engineering experience.
- Extensive knowledge of PyTorch, TensorFlow/Keras, ONNX, and TensorRT.
- Advanced Python proficiency with strong OOP, design, and problem-solving skills for large-scale applications, combined with familiarity and hands-on experience in Linux and Docker.
- Solid understanding of OpenCV and state-of-the-art DL algorithms for image classification, object detection, tracking, and segmentation.
- Well-versed in QA methodologies with a deep understanding of NVIDIA GPU technologies (e.g., RTX, DLSS).
- Hands-on experience building agentic gameplay systems using LLM/VLM-based agents, GenAI, RAG, vLLM, and solving complex problems with AIGC; proficient with AI development tools (Codex, Cursor, MCP, CodeRabbit) for test automation and workflow acceleration.
- Excellent written and verbal communication, strong initiative, self-motivation, and a commitment to high software quality standards.
Ways to stand out from the crowd:
- Knowledge of Transformer based LLM and AIGC, Imitation Learning, Model free/based RL, Hierarchical RL, Inverse RL, Meta-learning, Life-long learning.
- Hands-on experience in solving complex problems using Deep learning Algorithms would be a plus.
- Experience and medals in data science or computer vision competitions (e.g., Kaggle, CVPR workshop) will be a plus.
- Demonstrated ability to rapidly understand game dynamics and decompose game scenarios into solvable DL problems that can be implemented end‑to‑end.