Authoring repository-level context for AI coding agents: instruction files, skills, prompt libraries, and the conventions that make them consistent across many repositories.
• Reading and reasoning about unfamiliar production code in at least two languages well enough to describe what it does and why.
• Automated documentation and code-comprehension generation across legacy codebases.
• Context and retrieval design: what to include, what to exclude, chunking and indexing, grounding agent output in real repository facts.
• Robust technical writing for a developer audience, and the discipline to templatize rather than hand-craft each product.
• Evidence of measurably improving agent output quality by improving context, not by changing the model.

Nice To Have
Exposure to legacy stacks — COBOL in particular — where added application and domain context is what makes agentic work viable.
• AST, code-graph or static-analysis tooling used to generate context automatically.
• Information architecture or taxonomy background.

Responsibilities




Authoring repository-level context for AI coding agents: instruction files, skills, prompt libraries, and the conventions that make them consistent across many repositories.
• Reading and reasoning about unfamiliar production code in at least two languages well enough to describe what it does and why.
• Automated documentation and code-comprehension generation across legacy codebases.
• Context and retrieval design: what to include, what to exclude, chunking and indexing, grounding agent output in real repository facts.
• Strong technical writing for a developer audience, and the discipline to templatize rather than hand-craft each product.
• Evidence of measurably improving agent output quality by improving context, not by changing the model.

Qualifications
Nice To Have
Exposure to legacy stacks — COBOL in particular — where added application and domain context is what makes agentic work viable.
• AST, code-graph or static-analysis to

📌 AI Central (Pune)
🏢 Zensar
📍 Pune