Senior Machine Learning Engineer - Large Language Models
SoundHound
·
Posted:
August 16, 2023
·
Hybrid
About the position
SoundHound AI is seeking a candidate for a role focused on developing and implementing innovative neural architectures and pipelines for end-to-end Automatic Speech Recognition (ASR) and large language models. The ideal candidate should have a deep understanding of transformer models, optimization techniques for improving inference-time performance, and reinforcement learning techniques. Additionally, experience with distributed training and inference, as well as knowledge of statistical language models and probability theory, is highly desirable. The candidate should also have strong coding and implementation skills, particularly with PyTorch.
Responsibilities
- Develop and implement innovative neural architectures and pipelines for end-to-end ASR and large language models.
- Innovate in developing novel approaches and architectures for integration of large language models with end-to-end ASR to support prompting.
- Perform data science/analytics.
- Have strong foundations in AI/ML fundamentals.
- Have knowledge of neural networks and transformer models as applied to NLP.
- Have knowledge of classical NLP.
- Have knowledge of hyperparameter optimization methods such as Bayesian Optimization.
- Have a deep understanding of transformer models and associated architectures.
- Have knowledge of optimization techniques for improving the inference-time performance of transformer models.
- Have a deep understanding of GPT ("Generative Pre-Trained Transformer") models.
- Be familiar with reinforcement learning techniques in LLM models and instruction following training methods.
- Have knowledge of large language models, training validation, testing, and deployment.
- Be familiar with GPT-3, ChatGPT, and ChatGPT.
- Have knowledge of distributed training.
- Have knowledge of prompt engineering.
- Stay updated on the latest research developments in the field.
- Have the ability to develop efficient training pipelines using PyTorch.
- Have experience with distributed training and inference.
- Be familiar with associated libraries such as DeepSpeed.
- Have knowledge of ideal deployment methods and technologies and relative tradeoffs in MLOps.
- Have experience and interest in statistical language models such as n-gram models.
- Have strong foundations in probability theory and statistics.
- Understand concepts such as bias/variance tradeoff, model generalization, and bootstrapping methods.
- Have coding and implementation skills.
Requirements
- Strong foundations in AI/ML fundamentals
- Knowledge of neural networks and transformer models, particularly in NLP
- Familiarity with classical NLP
- Experience with hyperparameter optimization methods, such as Bayesian Optimization
- Deep understanding of end-to-end ASR and transformer models
- Knowledge of optimization techniques for improving inference-time performance of transformer models
- Familiarity with GPT models and reinforcement learning techniques in LLM models
- Experience with prompt engineering
- Proficiency in training, validation, testing, and deployment of large language models (GPT-3, ChatGPT)
- Knowledge of distributed training and associated techniques
- Experience with PyTorch (TensorFlow knowledge is desirable)
- Familiarity with libraries such as DeepSpeed
- Understanding of MLOps and deployment methods for ML models
- Experience and interest in statistical language models (SLMs)
- Strong foundations in probability theory and statistics
- Coding and implementation skills, demonstrated proficiency with P
Benefits
- Comprehensive health care
- Paid time off
- Discounted fitness classes
- Flexible working hours