Software Engineer, CUDA Deep Learning Systems

Jobtailor; California (MO); On-site; USD 140,000 - 210,000; Full time

Job summary

Jobtailor is seeking a highly skilled CUDA-focused engineer to optimize deep learning workloads across single-node to cluster-scale systems.

You will design custom CUDA kernels, improve performance for transformer-based models, and collaborate with researchers and kernel experts to push the boundaries of AI acceleration.

Qualifications

• BS/MS/PhD in CS/CE/EE or related field, or equivalent experience.
• 2+ years of industry or academic experience after degree.
• Proficiency in C++ and Python.
• Strong fundamentals in deep learning, especially transformers.
• Understanding of distributed computing and multi-node scaling.
• Experience in systems programming and low-level performance optimization.
• Familiarity with GPU DL accelerator architectures.
• Hands-on CUDA programming, kernel optimization, and profiling.
• Experience with generative AI models and LLMs.
• Research background in ML systems or related fields.

Responsibilities

• Explore, research, and prototype systems optimizations for deep learning models and CUDA.
• Architect and optimize distributed computing systems from single-node to cluster-scale environments.
• Design and optimize custom high-performance CUDA kernels for new neural architectures.
• Analyze hardware-software interactions to resolve bottlenecks in training and inference.
• Collaborate with researchers, architects, kernel authors, and CUDA experts to co-design systems.
• Develop tools and runtimes to profile and accelerate new deep learning paradigms.
• Write clean, maintainable code and transition prototypes into open-source releases or products.

Skills

CUDA Programming

C++

Python

Deep Learning Fundamentals

Distributed Computing

Computer Architecture

Kernel Optimization

Workload Profiling

Generative AI Models

Transformers

Education

BS/MS/PhD in CS/CE/EE or equivalent

Tools

PyTorch

JAX

TensorRT

NCCL

MPI

UCX

Triton

XLA

Torch.compile

SgLang

Job description

• Explore, research, and prototype systems optimizations for advanced deep learning models at the intersection of high-level deep learning frameworks and low-level CUDA.
• Architect and optimize distributed computing systems from single-node to cluster-scale supercomputing environments.
• Design, implement, and optimize custom high-performance CUDA kernels for emerging neural network architectures and workloads.
• Analyze hardware-software interactions to identify and resolve performance bottlenecks in training and inference pipelines.
• Collaborate with AI researchers, hardware and software architects, kernel and compiler authors, and CUDA driver experts to co-design systems and algorithms.
• Develop exploratory tools and runtime systems to profile and accelerate new deep learning paradigms.
• Write clean, effective, and maintainable code and transition prototypes into open-source releases, framework integrations, internal tools, or commercial products.

Requirements

• BS, MS, or PhD degree in Computer Science, Computer Engineering, Electrical Engineering, or related field, or equivalent experience.
• 2+ years of relevant industry experience or equivalent academic experience after degree achievement.
• Strong proficiency in C++ and Python programming.
• Solid background in deep learning fundamentals, focused on transformers.
• Strong understanding of distributed computing, multi-node scaling, and cluster-scale performance challenges.
• Proven experience in systems programming, computer architecture, and low-level systems performance optimization.
• Familiarity with GPU deep learning accelerator architectures.
• Hands-on experience with CUDA programming, kernel optimization, and workload profiling.
• Experience profiling and optimizing generative AI models, including large language models.
• Research background in machine learning systems or adjacent fields.
• Experience profiling and optimizing innovative vision models, generative AI architectures, or diffusion models.
• Track record of initiative and willingness to deep-dive on problems across the stack.
• Preferred experience with PyTorch, JAX, TensorRT, vLLM, sgLang, Nemo, or Megatron internals and execution graphs.
• Preferred hands-on experience with NCCL, MPI, or UCX and distributed machine learning techniques.
• Preferred knowledge of numerical methods and low-precision arithmetic such as NVFP4, MXFP4, FP8, and INT8.
• Preferred background in deep learning compilers and ML systems, including Triton, XLA, and torch.compile.
• Preferred experience with highly parallel or reinforcement-learning-style simulation environments.
• Preferred experience designing and implementing agentic AI systems for complex systems and infrastructure problems.

Core Competencies

Demonstrates expertise in CUDA programming, high-performance computing, and deep learning optimization, with a strong foundation in distributed systems and machine learning architectures. Capable of collaborating with cross-functional teams to design and implement innovative solutions for advanced AI models.

Highest-signal resume keywords

• CUDA Programming
• C++ and Python Proficiency
• Deep Learning Optimization
• Distributed Computing Systems
• Performance Bottleneck Analysis

ATS Optimization Keywords

Hard Skills

• CUDA
• C++
• Python
• Deep Learning Fundamentals
• Systems Programming
• Computer Architecture
• Kernel Optimization
• Workload Profiling
• Generative AI Models
• Transformers

Soft Skills

• Collaboration
• Initiative
• Problem-Solving

Industry Keywords

• Machine Learning Systems
• High-Performance Computing
• AI Research
• Neural Network Architectures
• Cluster-Scale Performance

Tools & Technologies

• PyTorch
• JAX
• TensorRT
• NCCL
• MPI
• UCX
• Triton
• XLA
• Torch.compile
• SgLang