Language Model Scientist

We're hiring a Language Model Scientist in Santa Clara, California, to join L&T's AI Center of Excellence. You'll build the generative and agentic AI systems that power knowledge, engineering and operations across the L&T Group, turning foundation-model capabilities into secure, reliable and domain-aware solutions: copilots, AI agents, enterprise knowledge systems and natural-language interfaces to engineering and physical systems.

• Fine-tune and align open-source and commercial LLMs (SFT, LoRA/PEFT, RLHF, DPO) for domain tasks such as engineering specifications, contracts, standards and code.
• Design agentic systems with tool use, planning, memory and multi-agent orchestration, plus RAG and knowledge-graph systems over enterprise data.
• Build evaluation, guardrail and red-teaming frameworks covering accuracy, hallucination, bias and safety.
• Deploy and scale LLMs in production (vLLM, TensorRT-LLM, quantization), with monitoring and inference cost optimization across cloud and on-premise.
• Partner with businesses in Engineering, Manufacturing, Infrastructure, Energy, Financial Services, IT Services and Defence to deliver high-value GenAI use cases.

• Enterprise LLM and agentic AI reference architecture defined for the CoE.
• 5–8 GenAI and agentic use cases delivered, at least 3 scaled to production.
• At least one domain-adapted language model developed.
• LLM evaluation, guardrail and LLMOps framework established.

• Master's or PhD in Computer Science, AI, NLP, Computational Linguistics or a related field (PhD preferred).
• 8–15 years in ML or NLP, including 3+ years of hands‑on work with LLMs and Generative AI.
• A track record of building and deploying LLM applications in production at enterprise scale.
• Deep knowledge of transformer architectures, pre‑training, fine‑tuning and alignment.
• Strong Python with PyTorch, Hugging Face, Deep Speed/FSDP and LLM frameworks such as Lang Chain, Llama Index or Lang Graph.
• Experience with vector databases, knowledge graphs, Model Context Protocol (MCP) and function calling.

Plus: engineering, industrial or regulated-domain experience, and publications at ACL, EMNLP, NAACL, NeurIPS, ICML or ICLR.

Role details:
Reports to the Head, AI Center of Excellence. Headcount:
3. Scope: L&T Group-wide, United States and India.