Developer Advocate – Distributed Training

NVIDIA•Us, CA

5d•Hybrid

About The Position

The NVIDIA developer marketing group is looking for a developer advocate who is passionate about working with innovative technology, building well-documented inspirational demos, and sharing that code with the AI community. Our team is deeply committed to championing the developer ecosystem that builds applications on NVIDIA’s platform. We need a candidate who understands developers and works with the developer marketing on launching new technology. We need hard-working and creative people who are passionate about teaching our developer community how to employ accelerated computing.

Requirements

World-class communication and presentation skills with a proven track record of articulating the value proposition of an emerging technology.
Experience presenting to technical audiences and writing developer-facing content.
Familiarity with all kinds of scaling laws and experience in optimizing training efficiency.
Strong programming skills with the ability to develop easy-to-understand examples for training pipelines using AI frameworks such as PyTorch, JAX, DeepSpeed, and NeMo.
A Bachelor's degree from a leading university in Computer Science, or equivalent experience, along with 3 years of engineering experience or equivalent.

Nice To Haves

Master's or Ph.D. degree from a leading university in software engineering or computer science.
In-depth knowledge of software architecture, system design, and application development.
Hands-on experience with popular training libraries such as MaxDiffusion, OpenMMLab, HuggingFace Accelerate, Megatron.

Responsibilities

Create impactful and relevant content (e.g., demos, blogs, presentations, videos, etc.) and engage with developers, researchers, and students at events online or in person.
Work with the product marketing team to launch emerging technology, write code, and share it with the community to inspire developers.
Be active on public-facing NVIDIA channels, including forums, Discord, and social media, helping developers understand our demos and encouraging them to build their own applications.
Design and implement distributed learning and post-training pipelines for LLMs/VLMs to improve model reasoning, safety, and instruction-following capabilities.
Optimize training efficiency.
Develop easy-to-understand examples for training pipelines using AI frameworks such as PyTorch, JAX, DeepSpeed, and NeMo.
Prioritize and work through constantly evolving projects that strive to demonstrate the benefits of new, potentially unstable technology.