Applied Scientist

We are hiring an Applied Scientist with experience in performing hyperparameter optimizations, evaluating model performance, and deploying production natural language systems. Working with an elite group of professions, you will have the opportunity to work hands on with ML research and development. This environment allows for vigorous dialogue on the problems needed to solve for and the opportunity to share bleeding edge solutions. Ultimately though, the role requires an individual who can commit to team conclusions as you work towards a common goal.

A skilled researcher, the successful candidate will develop novel methods to obfuscate data for common NLP tasks such as question/answering, instruction following, and sentiment analysis to enhance cutting-edge LLMs to preserve user privacy effectively. You will directly interface with Large Language Models (such as GPT, BLOOM, LaMDA, and LLaMA) and also with their core NLP privacy technology. You will also develop tools and metrics for handling machine learning experiments during training, inference, logging, and analytics. The successful candidate will also bring a deep understanding of transformer architecture, adversarial training knowledge.

Required:

• Experience with deploying distributed systems for training deep learning models
• Experience with developing NLP systems, specifically with language modeling.
• Experience with fine-tuning Large Language Models (LLM) on external datasets.
• Hands-on experience with using HuggingFace APIs.
• 4 Years Experience with Python and 2 Years of Professional Experience with PyTorch, NumPy

Required

• Ph.D./MS in Computer Science or Electrical and Computer Engineering, specializing in NLP and deep learning.
• Speed optimization/ optimization of memory consumption for LLM
• Proven ability to impact products with cutting-edge research technology.
• Experience with downstream NLP models (includes question answering, sequ
• ence labeling, and downstream classification.)
• Experience with using Kubernetes.
• Experience training LLM’s such as GPT, BERT, LaMDA, and LLaMA.
• Comfort with fast R&D cycles (ideation through to deployment and production systems).
• Familiar with concepts in data privacy, such as differential privacy, secure multi-party computation, NLP red teaming, and homomorphic encryption.