Number of Position(s): 1 Duration: 10 Weeks Date: June 2026 to August 2026 Location: Hybrid, in Murray Hill, NJ. EDUCATIONAL RECOMMENDATIONS - Currently a candidate for a PhD in Computer Science, Computer Systems Engineering, Math, Artificial Intelligence, or related field at an accredited school in the USA. Design and implement state-of-the-art AI/ ML decentralized systems. Validate and evaluate your implementation in our cutting-edge labs. Interface, explore, and learn from the experts. Expertise in deep learning fundamentals, including large language models and agent-based systems, and experience with training, deploying, and/or profiling models. Experience in principled systems design and development. Excellent communication skills, with the ability to analyze complex problems and effectively communicate findings. Strong publication record in top-tier AI and systems conferences. It would be nice if you also had:We encourage applications from candidates who have a strong foundation in one or more of the areas below, even if you don’t meet every criterion. We value diverse perspectives, innovative thinking, and complementary skills. Agentic AI & Large Language Models (LL - Ms):Familiarity with large-scale model inference and optimization, as well as experience in LLM reasoning, prompt engineering, and resource-constrained computation. AI Systems Architecture & Optimization:Experience managing GPU or accelerator resources, optimizing performance, and benchmarking across different hardware environments. A solid understanding of AI infrastructure design and inference workflows—such as KV-cache management, batching, and offloading—is beneficial. Compilers & Hardware–Software Co-Design:Knowledge of computational graph representations (e.g., ONNX, MLIR, XLA, Torch. Script) and model optimization frameworks (e.g., Tensor. RT, TVM). Experience working with heterogeneous accelerator ecosystems (e.g., TP - Us, AMD RO - Cm GP - Us) or parallelizing compilers is a plus. Distributed, Edge AI & Web 3 Computing:Understanding of distributed or edge inference systems (e.g., Ray Serve, Deep. Speed-Inference, v. LLM), with familiarity in blockchain technologies, smart contracts, or wireless networking protocols (Wi-Fi, 3 GPP, Bluetooth).