ML Researcher, Foundational Models

Highlights

• Drive open-ended research that shapes Sarvam's next generation foundational models.
• Make architecture and training decisions on frontier-scale runs with large compute.
• Own research proposals through to production alongside infra and data teams.

Ideal candidate

Curious, opinionated researcher with PhD-level depth who loves open-ended questions, rigorous ablations, and turning hunches into shipped frontier models with real impact for India.

Requirements

• PhD in ML, Computer Science or related field (or in final stages of completion)
• 3+ years of research experience post-PhD or equivalent depth
• First-author publications at top-tier venues like NeurIPS, ICML, ICLR, ACL, EMNLP, COLM
• Hands-on experience pre-training transformer LMs from scratch, ideally 7B+ parameters end-to-end
• Meaningful contributions to open-source LLM ecosystem (code, models, datasets)
• Fluency in PyTorch and comfort with distributed training
• Strong intuition for experimental design, ablations and scaling behaviour