AI Research Scientist/Engineer

Menlo Park, CA | On-Site | Full-Time/Direct Hire 

Seeking top-tier PhDs (Bay Area preferred) with ICML/ICLR publications in LLM training and inference optimization—no vision/audio, just pure language; diffusion model experience a plus.

Client Opportunity | Through Phizenix (WBENC & Minority-Certified Recruiting Partner)

Join a trailblazing AI startup that's reinventing how large language models are built—with diffusion-powered LLMs that generate faster, adapt smarter, and handle multimodal data like no other.

We’re looking for a Research Scientist / Engineer who’s ready to move beyond traditional autoregressive methods and help shape the next wave of generative AI. You'll collaborate with pioneers in AI research, design novel model architectures, and scale your ideas from paper to production.

• Design and refine LLM architectures built on a diffusion-first paradigm

• Develop cutting-edge training strategies and custom loss functions

• Translate research into real-world systems for enterprise-scale deployments

• Explore constraint-aware generation and controlled outputs

• Push the limits of model efficiency, scalability, and multi-modal capabilities

• Must have a PhD in Computer Science, Machine Learning, or a related field

• Hands-on experience with PyTorch and LLM fundamentals (transformers, KV caching, etc.)

• Should have recent/or any ICLR/ICML publications in LLM inference optimization would be ideal. 

• Familiarity with diffusion models and distributed model training

• Solid research-to-production mindset with 2+ years in an ML/AI role

• Training LLMs from scratch and optimizing large-scale runs

• Advanced training tactics (e.g., mixed precision, gradient accumulation)

• Experience with cross-modal modeling and inference frameworks like vLLM, TensorRT

• A background in model efficiency, optimization theory, or infrastructure-aware research

At Phizenix, we’re proud to partner with diverse and inclusive teams building AI that matters. If you're ready to build, innovate, and publish with one of the boldest teams in AI—this is your moment.