Senior DL Software Engineer, Model Optimization and Edge Deployment - Autonomous Vehicles

Overview

NVIDIA is a leader in the AI field, specifically in Embodied AI and autonomous vehicles. This role is for a Deep Learning Engineer who will bridge cutting‑edge multimodal architectures with real‑time robotic execution, ensuring large language models (LLM) and vision‑language models (VLM) run fast, lean, and reliably on on‑board hardware.

• Develop state‑of‑the‑art model optimization techniques such as speculative decoding, KV cache streaming, and prefill‑decode separation to boost end‑to‑end performance for production deployments.
• Implement advanced compression methods, including Quantization (FP4/FP8), pruning, and knowledge distillation, to reduce model footprints without sacrificing safety‑critical accuracy.
• Design high‑performance inference strategies, including automated model sharding (tensor/sequence parallelism) and efficient attention kernels optimized for KV‑caching.
• Conduct detailed layer‑by‑layer profiling to identify compute and memory bottlenecks, driving targeted optimizations for real‑time execution.
• Utilize the PyTorch ecosystem to extract standardized model graph representations and automate deployment pipelines for Tensor
  
  RT conversion.
• Scale deep learning model performance across NVIDIA’s edge architectures, optimizing throughput on specialized accelerators.
• Architect a clean software interface that integrates large‑scale models within a high‑performance C++ production environment.
• Collaborate with research, Tensor
  
  RT, and Cosmos teams to translate breakthroughs into shipping product solutions.

• PhD with 4+ years, MS with 6+ years, or BS (or equivalent) with 8+ years of experience in Computer Science, Computer Engineering, or a related technical field.
• Expert proficiency in PyTorch, JAX, or similar machine‑learning frameworks.
• Sophisticated experience with modern LLM/VLM inference stacks such as vLLM, Tensor
  
  RT‑LLM, and SGLang.
• Proven track record of training, deploying, or optimizing large‑scale DL models in production environments.
• Deep familiarity with NVIDIA’s deep‑learning SDKs, especially Tensor
  
  RT and CUDA.
• Strong understanding of GPU architecture, compilation stacks, and the ability to debug end‑to‑end performance across hardware and software boundaries.

• Deep experience with LLM, VLM, and VLA model optimization tailored for real‑time robotic control, embodied AI, and autonomous decision‑making.
• Track record of implementing low‑bit inference and writing custom high‑performance kernels using CUDA, Triton, or CUTLASS to accelerate non‑standard neural‑network layers.
• Active contributions to open‑source inference and optimization libraries such as vLLM, SGLang, and Tensor
  
  RT‑LLM.
• Comprehensive understanding of real‑time robotics constraints, including safety‑critical determinism, hardware‑in‑the‑loop testing, and ultra‑low latency requirements.

Base salary ranges from USD
184,000–287,500 for Level4 and USD
224,000–356,500 for Level5, depending on location, experience, and comparable market salaries. Eligible for equity and standard NVIDIA benefits.

NVIDIA is an equal‑opportunity employer and does not discriminate on the basis of race, religion, color, national origin, gender, gender expression, sexual orientation, age, marital status, veteran status, disability status, or any other characteristic protected by law. NVIDIA is committed to fostering an inclusive work environment.