Senior Gen AI Engineer

You will partner with business stakeholders to design AI-driven extraction rules that convert unstructured documents into high-value structured data. You'll craft and evaluate prompts for LLM-based extraction, maintain a Python codebase that powers workflows, and translate stakeholder needs into repeatable, auditable extraction pipelines.

Responsibilities:

• Translate business requirements into precise extraction targets, schemas, and quality thresholds.
• Design, test, and iterate prompts for LLM-based extraction; implement evaluation protocols (e.g., accuracy, precision/recall).
• Configure and maintain extraction workflows in Python, including parsing, validation, and post-processing rules.
• Triage and resolve minor code issues; write unit tests and improve logging, monitoring, and error handling.
• Work closely with the GB Data Science team in the US in an onsite-offshore model.
• Partner with business owners to review outputs, prioritize improvements, and manage a backlog in Azure DevOps.

About you:

Preferred qualifications:

• An undergraduate/Masters degree in Computer Science, Data Science, Information Systems, or equivalent experience.
• Experience with enterprise-scale datasets or high-volume document processing.
• Exposure to Azure services, ideally Azure AI Search/Cognitive Search and Azure OpenAI, Databricks, Snowflake.
• Employees should have experience in Python/Spark, CI/CD, Scikit-learn, Git.
• Practical experience crafting prompts for LLMs and evaluating outputs (few-shot prompts, structured output formats, guardrails).
• Ability to read/modify Python scripts, debug minor issues, and write clear, maintainable code.
• Strong communication skills: can elicit requirements, clarify edge cases, and document decisions and workflows.
• Working knowledge of handling sensitive data securely and following standardized procedures.

Experience: 2-5 years as AI Engineer or Prompt Engineer, understanding of designing data science experiments

Skills

Azure, Databricks, LLM, Prompts, Python, Snowflake