Machine Learning Engineer (LLM/Speech Processing)

Responsibilities:

• Develop and Implement Advanced Speech Processing Models: Design, implement, and optimize machine learning models for TTS and forced alignment, with a strong emphasis on phoneme-level precision. Experiment with and implement models like Wav2Vec2, and other relevant speech processing architectures.
• Leverage Hugging Face and Open-Source Tools: Utilize the Hugging Face ecosystem for model development, training, and deployment.
• Optimize for GPU Performance: Develop and optimize models for efficient execution on GPU hardware. Profile and debug performance bottlenecks to ensure optimal training and inference speeds.
• Data Processing and Feature Engineering: Process and prepare large datasets for training and evaluation. Develop and implement feature engineering techniques to improve model performance 
• Research and Development: Stay up-to-date with the latest advancements in speech processing and LLMs. Conduct research to explore novel approaches and improve existing techniques. Implement and evaluate new algorithms and techniques.
• Collaboration and Communication: Work closely with other engineers and researchers to achieve project goals. Communicate technical findings and progress effectively. Document code and methodologies thoroughly.
• Forced Alignment implementation: Implement and improve forced alignment algorithms, with the goal of high accuracy phoneme level timestamps.

Requirements:

• Strong Background in Machine Learning: Proven experience in developing and training machine learning models. Solid understanding of deep learning concepts and techniques.
• Experience with Speech Processing: Hands-on experience with speech processing tasks, such as TTS, forced alignment, or speech recognition. Familiarity with acoustic modeling and phonetics. Experience with Wav2Vec2 or similar speech models.
• Expertise in LLMs: Experience working with Large Language Models (LLMs) and their applications. Understanding of LLM architectures and training techniques.
• Proficiency in Python and Deep Learning Frameworks: Strong programming skills in Python. Proficiency in deep learning frameworks like PyTorch or TensorFlow.
• Experience with Hugging Face: Practical experience using the Hugging Face Transformers library and ecosystem.
• GPU Optimization: Experience optimizing machine learning models for GPU acceleration. Familiarity with CUDA and related technologies.
• Open-Source Contribution: Experience contributing to open-source projects is a plus.
• Strong Problem-Solving and Analytical Skills: Ability to analyze complex problems and develop effective solutions.
• Excellent Communication and Collaboration Skills: Ability to work effectively in a team environment.