Context Trainer (Developer)

By clicking the “Apply” button, I understand that my employment application process with Takeda will commence and that the information I provide in my application will be processed in line with Takeda’s Privacy Notice and Terms of Use . I further attest that all information I submit in my employment application is true to the best of my knowledge.

Job Description THE OPPORTUNITY Curate and govern the context layer (RAG/KBs, embeddings, metadata, labeling) to improve answer quality and minimize hallucinations, while protecting data/PII.

RESPONSIBILITIES

Curation & Labeling Extract and curate content from enterprise sources (Confluence, Jira, SharePoint, ServiceNow, qTest ) using APIs and automation. Define chunking and metadata schemas; labeling guidelines; golden Q&A and evaluation sets. Implement chunking strategies for diverse content types (code repositories, technical documentation, tickets, test cases).

Implement curation workflows and retention policies. Retrieval Quality Run A/B experiments across vector stores; monitor answer quality vs. cost/latency; recommend defaults. Analyze failure cases and propose data-driven improvements. Data Governance Enforce data minimization, retention, and access controls; maintain lineage and approvals per RAI (Responsible AI).

Document data sources and usage for audit readiness. SKILLS & QUALIFICATIONS Required 3 + years data/ML experience with embeddings/retrieval expertise ; strong documentation and runbook skills. Experience with content transformation, metadata extraction, and labeling workflows. Familiarity with privacy and data governance principles.

Hands-on experience with vector stores (OpenSearch/ pgvector /Kendra/Chroma) and labeling tools. Experience with REST APIs and data extraction from enterprise systems. Python coding proficiency for data pipelines and automation. Preferred/Nice to have Experience designing golden datasets and evaluation pipelines. AWS Bedrock Knowledge Bases experience.

Familiarity with software development lifecycle and technical documentation patterns. Locations IND

• Bengaluru Worker Type Employee Worker Sub-Type Regular Time Type Full time