Senior AI Engineer: Agentic LLM Pipelines Banking

Job Description

This is a 12-month extendable contract supporting an AI Centre of Excellence, including agentic AI systems, Lang Graph workflows, LLM pipelines, HITL mechanisms, AI safety guardrails, Azure AI services, and banking use cases.

Role Overview

• The AI Engineer role is a hands-on, production-focused position responsible for building and orchestrating advanced agentic AI systems, initially for Intelligent Document Processing (IDP) in cheque clearing and scaling across banking use cases.
• The role involves designing Lang Graph-based agent workflows, developing LLM-powered extraction pipelines with strong schema validation and prompt governance, and integrating MCP tools and Azure AI services into a dual orchestration stack.
• It also includes implementing human-in-the-loop mechanisms, AI safety guardrails, and rigorous model evaluation frameworks.
• Ideal candidates should bring strong Python expertise, experience with agentic AI frameworks, prompt engineering, APIs, Azure-based AI platforms, structured debugging of non-deterministic systems, and an interest in banking domain processes.

Build agentic AI systems; LLM pipelines and prompts; HITL and guardrails.

• The Senior AI Engineer role is a leadership-level, end-to-end technical position responsible for delivering complete AI use cases from design through production.
• It involves owning the architecture of complex multi-agent Lang Graph systems, defining evaluation frameworks, and calibrating model confidence and human-in-the-loop thresholds in collaboration with business stakeholders.
• The role places strong emphasis on LLM evaluation, including metrics such as F1, precision, recall, calibration, regression testing, and ensuring production reliability through structured CI/CD pipelines.
• In addition to hands-on technical design across RAG, orchestration, and Azure AI platforms, the Senior AI Engineer mentors junior engineers, leads code reviews, drives adoption of new AI capabilities, and proactively identifies and mitigates failure modes before deployment.

Own AI architecture delivery; LLM evaluation and calibration; mentor engineers and prevent failures.