The Role
Join Luma's applied research team as a Senior Research Engineer, working on multimodal AI systems. Requires strong Python skills, experience in visual AI, and a passion for generative AI. Must have experience with cluster orchestration and large-scale ML systems. This role is not for recent grads.
Summary Generated by Built In
We are working on training and scaling up multimodal foundation models for systems that can see and understand, show and explain, and eventually interact with our world to effect change.
We are looking for people who mix strong technical ML skills, boundless curiosity and a constant desire to make better products. You’ll be part of Luma’s applied research team and work directly on mission critical work streams utilizing thousands of GPUs.
Responsibilities
- Implement cutting-edge generative AI features for inclusion in our product
- Work with Research and Product to fine-tune models, including dataset creation, model fine-tuning and evaluation
- Build tools & methods for evaluating our models and identify & implement solutions
- Train and run ML classifiers & embeddings to categorize by data quality and other attributes
Experience
- Strong generalist Python skills including significant experience with PyTorch.
- Recent experience in, and understanding of, visual AI, including diffusion models and transformers (note: can be from professional work, academic research experience, or hobby work with open source models like Stable Diffusion)
- Passion for experimenting with multi-modal generative AI, including familiarity with recent research work and ideas for implementing research into usable products.
- Good to have experience with cluster orchestration tools like SLURM
- Good to have experience with high performance large scale ML systems (>100 GPUs)
- Please note this role is not meant for recent grads.
Your application is reviewed by real people.
Top Skills
Python
The Company
What We Do
Luma is a multimedia platform that delivers personalized movie and TV program selections from a range of sources to its viewers.