Research Staff, LLMs [EMEA - Remote]

The role

This researcher will define new LLM research initiatives and design experimental programs for LLMs. You will drive transformer training jobs on distributed compute infrastructure and deploy new models into production.

What you will do

• Brainstorming and collaborating with other members of the Research Staff to define new LLM research initiatives
• Broad surveying of literature, evaluating, classifying, and distilling current methods
• Designing and carrying out experimental programs for LLMs
• Driving transformer (LLM) training jobs successfully on distributed compute infrastructure and deploying new models into production
• Documenting and presenting results and complex technical concepts clearly for a target audience
• Staying up to date with the latest advances in deep learning and LLMs

Required qualifications

• 3+ years of experience in applied deep learning research
• Solid understanding toward the applications and implications of different neural network types, architectures, and loss mechanism
• Proven experience working with large language models (LLMs) including data curation, distributed large-scale training, optimization of transformer architecture, and RL Learning
• Strong experience coding in Python and working with Pytorch
• Experience with various transformer architectures (auto-regressive, sequence-to-sequence.etc)
• Experience with distributed computing and large-scale data processing
• Prior experience in conducting experimental programs and using results to optimize models

Nice to have

• Deep understanding of transformers, causal LMs, and their underlying architecture
• Understanding of distributed training and distributed inference schemes for LLMs
• Familiarity with RLHF labeling and training pipelines
• Up-to-date knowledge of recent LLM techniques and developments
• Published papers in Deep Learning Research, particularly related to LLMs and deep neural networks

About the company

Deepgram is a research company that productizes as fast as possible. Their speech models and real-time infrastructure power AI systems for more than 5,000 organizations.