Engineer, Principal

Responsibilities

• Optimize latest LLM and GenAI model architectures for NPUs, including reimplementing basic building blocks.
• Train and fine-tune Large Language Models (LLMs) from scratch and on specific datasets using state-of-the-art techniques.
• Implement advanced training methods such as Reinforcement Learning from Human Feedback (RLHF), ZeRO, and Speculative Sampling.
• Manage large datasets, ensuring data quality through effective cleaning, preprocessing, and Exploratory Data Analysis (EDA).
• Evaluate model performance using metrics such as MMLU, Rouge, BLEU, and Perplexity.
• Develop infrastructure to optimize ONNX memory footprint and export time.
• Stay current with the latest NLP and LLM research to contribute to ongoing R&D efforts.

Requirements

• Bachelor's degree in Computer Science, Machine Learning, AI, or a related field.
• 4+ years of software engineering experience (or 3+ years with a Master's, or 2+ years with a PhD).
• Hands-on experience with PyTorch at a granular level, including tensor operations and GPU acceleration.
• Strong understanding of Natural Language Processing (NLP) and Large Language Models (LLMs).
• Proficiency in Python and experience with software development best practices.
• Proven experience working with large-scale datasets.
• Experience with programming languages such as C, C++, or Java.

Preferred Qualifications

• 9+ to 12 years of relevant industry experience.
• Knowledge of quantization techniques (AWQ) and Quantization Aware Training (QAT).
• Experience with Neural Architecture Search (NAS) techniques.
• Hands-on experience with CUDA, CUDNN, and Triton-lang.
• Familiarity with data version control tools.

About the Company

Qualcomm India Private Limited is part of the Qualcomm Cloud AI team, which develops cutting-edge hardware and software for Machine Learning solutions. Our work spans the data center, edge, infrastructure, and automotive markets, driving innovation across the global technology landscape.