Legal AI Model Alignment and Quality Assurance Specialist

About the Role

Join TathyaGrid's founding team to lead the creation, alignment, and quality control of bespoke legal AI models tailored for the Indian legal ecosystem. This role involves curating expert-level legal training datasets, auditing AI outputs for accuracy, and developing comprehensive evaluation frameworks.

Key Responsibilities

• Develop high-quality training datasets, instructional data, and fine-tuning samples for specialized legal large language models (LLMs).
• Perform rigorous audits, evaluations, and grading of model-generated content including summaries, analyses, and search results to detect errors and improve output reliability.
• Create gold-standard evaluation data sets incorporating complex Indian case law, tribunal rulings, and statutory regulations.
• Work closely with machine learning engineers to optimize prompt engineering, system commands, and retrieval-augmented generation (RAG) workflows.
• Annotate, clean, and organize historical court judgments and legal briefs to support domain-specific model training.
• Systematically review search queries and model summaries against benchmark references to identify hallucinations and inaccuracies.
• Contribute to the development of high-quality prompt templates and instruction-based examples to enhance model alignment.

Required Qualifications and Skills

• Holder of an LLB or LLM degree from a recognized Indian university.
• At least two years of post-qualification experience in litigation, legal research, or legal annotation.
• Comprehensive expertise in Indian case law, civil and criminal procedures, and corporate law.
• Strong analytical abilities combined with a passion for advancing legal technology and AI model alignment, including reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO).
• Alternatively, current final-year law students pursuing a 3-year or 5-year LLB, or recent law graduates with excellent written English skills capable of synthesizing complex case details.
• Deep knowledge of legal principles, case taxonomies, and judgment structuring.
• Interest and eagerness to engage with AI models, prompt engineering techniques, and technology-driven legal workflows.

Application Details

Prospective candidates can apply directly by submitting the required application materials including CV and cover note. Applicants are requested to consent to data processing in accordance with the Digital Personal Data Protection Act, 2023. Following the company page to receive updates is encouraged. This is an onsite full-time position based in Sevenig, Rhineland-Palatinate, Germany.