Lead AI Alignment & Evaluation Scientist

Wipro Technologies; San Francisco (CA); On-site; USD 240,000 - 375,000; Full time

Benefits offered by this job

Medical and dental benefits

Disability insurance

Paid time off

Job summary

Wipro Limited is seeking a Lead Researcher for AI data alignment and evaluation. You will architect data pipelines, design proactive benchmarks, and publish novel methodologies for LLMs and physical AI. This role emphasizes scientific rigor and leadership in alignment science.

Responsibilities include technical leadership, collaboration with researchers, and developing frameworks for evaluation across open-weight models, with a strong publication track record expected.

Qualifications

• Advanced degree in a highly quantitative field (CS/AI/Math).
• Deep familiarity with modern alignment techniques (RLHF, DPO, RLAIF).
• Experience with standard evaluation harnesses (EleutherAI LM Eval, HELM, OpenAI Evals).
• Strong ability to architect data pipelines and conduct rigorous evaluation.

Responsibilities

• Design Alignment Methodologies: Architect LHoH and synthetic data generation, design pairwise rubrics and reward mechanisms.
• Build Proactive Benchmarks: Create novel benchmarks spanning agentic reasoning to physical AI.
• Decontaminate & Validate: Detect data contamination and ensure rigorous evaluation to prevent reward hacking.
• Research Publication: Conduct independent research and co-author papers on evaluation frameworks.
• Technical Leadership: Collaborate with domain researchers to turn ground-truth data into scoreable evaluation frameworks.

Skills

RLHF

DPO

RLAIF

Evaluation harnesses

Education

PhD or high-impact MSc in CS/AI/Math

Tools

PyTorch

HuggingFace

vLLM

Llama 3

Mistral

Qwen

Job description

Wipro Limited is seeking a Lead Researcher for AI data alignment and evaluation. You will architect data pipelines, design proactive benchmarks, and publish novel methodologies for LLMs and physical AI. This role emphasizes scientific rigor and leadership in alignment science.

Responsibilities include technical leadership, collaboration with researchers, and developing frameworks for evaluation across open-weight models, with a strong publication track record expected.