AI Research Engineer – Model Compression, Quantization

This is a fully remote position, open to applicants in India.

📋 Description

• Lead the charge in innovating model compression and efficient deployment for state-of-the-art multimodal AI systems, which encompass large language models (LLMs) and vision-language models (VLMs).

• Implement and enhance compression methodologies such as quantization, knowledge distillation, and pruning to optimize intricate multimodal architectures that incorporate text, images, and audio.

• Develop robust compression workflows, establish performance and fidelity benchmarks, and resolve bottlenecks encountered during production inference.

⛳️ Requirements

• A degree in Computer Science or a related discipline.

• Preferably a PhD in NLP, Machine Learning, or a related area, backed by a strong history in AI research and development (with notable publications in A* conferences).

• Proficiency in PyTorch deep learning frameworks or comparable frameworks.

• Practical experience with model quantization, including both Quantization-Aware Training (QAT) and Post-Training Quantization (PTQ).

• Research and practical experience in knowledge distillation aimed at condensing large models into smaller, more efficient versions.

• Research and practical experience in model pruning for the purpose of minimizing large models into more compact, efficient forms.

• Comprehensive understanding of neural network architectures and training methodologies, including transformers (e.g., LLMs, VLMs), backpropagation, optimization, and fine-tuning strategies.

• Familiarity with C++ is advantageous, particularly for the implementation of low-level quantization kernels or inference enhancements.

🏝️ Benefits

• Flexible working hours

• Professional development opportunities