Researcher - Efficient Architecture for LLMs

About the role

The researcher will propose and develop novel ideas to push the state of the art in large language models (LLMs). The role involves leading applied research projects on long‑context LLMs, delivering end‑to‑end experiments, publishing results, and mentoring a small team of researchers and engineers.

Key responsibilities

• Propose and develop novel research ideas to advance the state of the art in large language models (LLMs).
• Lead applied research projects focused on long‑context LLMs, including efficient architectures, data curation, and post‑training methods.
• Design and execute experiments, analyze results, and present findings to internal and external audiences.
• Publish research in top‑tier ML and NLP conferences and mentor a small team of researchers and engineers.
• Monitor advances across the LLM ecosystem and translate insights into actionable project directions.

Required profile

• Ph.D. in Computer Science, Software Engineering, or a related field.
• First‑author publications in top‑tier venues such as NeurIPS, ICML, ICLR, ACL, EMNLP, or NAACL.
• Deep understanding of NLP and LLM fundamentals with strong theoretical grounding in machine learning.
• Excellent oral and written communication skills in English.

Required skills

• Strong programming in Python.
• Expertise with PyTorch.
• Experience using Hugging Face Transformers and Accelerate.
• Familiarity with large‑scale training frameworks such as Megatron‑LM and inference stacks like vLLM.
• Knowledge of modern transformer models (e.g., Llama, Qwen) and linear‑attention models (e.g., Mamba, GLA).