AI Researcher (Distillation)

TL;DR

AI Researcher (Distillation): Design and evaluate model distillation techniques for large language models and specialized architectures with an accent on tradeoffs between model size, latency, memory, and accuracy. Focus on developing novel distillation approaches, running large-scale experiments and ablations, productionizing research outcomes, and publishing papers to top-tier venues like NeurIPS, ICML, ICLR.

Location: Remote (world)

Company

Series A startup developing efficient, high-performance AI models with a focus on real-world deployment.

What you will do

• Design and evaluate model distillation techniques including teacher-student training, self-distillation, and representation matching.
• Research tradeoffs in model size, latency, memory, and accuracy for LLMs and inference-constrained environments.
• Develop novel distillation methods for long-context or specialized architectures.
• Run large-scale experiments, perform rigorous analysis, and collaborate with engineers to productionize results.
• Write and submit research papers to top conferences and contribute to open-source projects.

Requirements

• Strong background in machine learning research.
• Hands-on experience with model distillation or related topics like compression, pruning, quantization.
• Publication experience in conferences, journals, or arXiv preprints.
• Solid understanding of deep learning fundamentals including optimization and training dynamics.
• Fluency in PyTorch and research-grade experimentation.
• Ability to clearly communicate research ideas, results, and limitations.

Nice to have

• Experience distilling large language models.
• Work on efficiency-focused research like latency and throughput optimization.
• Experience with long-context models or non-Transformer architectures.
• Open-source contributions in ML or research tooling.
• Prior startup or applied research experience.

Culture & Benefits

• Real ownership over research direction.
• Strong support for publishing and open research.
• Tight feedback loop between research and deployment.
• Access to significant compute resources and production-scale problems.
• Small, highly technical team with deep ML and systems expertise.