Agentic AI Analyst

Responsible for building production-ready backend services for agentic workflow components aligned to solution architecture and platform standards. Implement solution designs by building Python services, worker processes, and reusable libraries following defined architecture, patterns, and standards. Develop agentic workflow components: tool connectors, orchestration steps, state management modules, retrieval components, and approval/escalation flows. Build reliable LLM interaction layers: tool/function calling, schema-validated structured outputs, guardrails, safe tool execution boundaries, and fallback behaviours. Implement robust backend patterns: async execution, job queues, retries/idempotency, compensating actions, and failure isolation for long-running workflows. Deliver production readiness: logging/tracing, metrics, decision logs, run replay support, performance profiling, and cost/latency controls. Write clean, maintainable, testable code with strong review discipline: unit/integration tests and regression testing for prompts/agents where applicable. Collaborate closely with the Agentic AI Architect and technical leads; support delivery across DEV/UAT/PROD including defect triage and operational support.

The role supports one of our top-tier banking clients in London (Canary Wharf) and requires a minimum of three days on-site presence. This is a permanent position based in the UK. We will only consider applicants who are eligible to work in the UK. For this role do NOT offer visa sponsorship.

Experience Requirements & Qualifications

• Core Experience
• 4+ years in Python backend development, including building production APIs/services and/or worker-based processing systems.
• Demonstrable experience in implementing Generative AI, AI/LLM-enabled features or systems (agentic workflows, RAG, tool calling, evaluation/monitoring) is preferred.
• Strong capability in backend fundamentals: service boundaries, API contracts, async execution, retries/idempotency, error handling, and performance optimization.
• Advanced Python engineering skills: clean architecture, modularity, testability, packaging, secure coding, and maintainability at team scale.
• Strong experience building API-first services (FastAPI or equivalent), RESTful APIs including auth patterns (OAuth2/JWT/API keys), versioning, and backwards compatibility.
• Integrate and manage relational and vector databases.
• Strong schema/data contract practice using typed models and validation (e.g., Pydantic-style patterns), including strict structured outputs and schema evolution.
• Working with version control tools like GitHub (branching, PR reviews, release tagging, CI-friendly workflows).
• Strong experience with context grounding methods, and context engineering when working with LLMs (RAG, evidence capture, context selection, prompt/context structuring).
• Experience using automation tools and integrating with external applications (API-based integrations, workflow triggers/actions, third-party systems).
• Experience building integration-heavy systems: consuming/producing APIs, handling enterprise data formats, and creating maintainable connectors.
• Working knowledge of distributed execution patterns: background jobs, scheduling, worker pools, and stateful workflows.
• Ability to work with ambiguity, break down requirements, and deliver reliably with strong ownership and communication.
• Nice to Have
• Experience with agent orchestration frameworks (e.g., LangGraph-like patterns) and LLM observability/evaluation tools (Langfuse-like capabilities).
• Experience integrating enterprise-hosted LLMs (including vertex AI / managed equivalents) and working with provider-agnostic abstraction layers (routing, fallback, cost-aware selection).
• Experience with job queues, distributed tracing, dashboards/alerts, and runbook-driven operational practices.
• Experience supporting regulated enterprise delivery: audit-friendly logging, change controls, secure configuration, and controlled deployments.
• Platform/DevOps awareness (preference): Docker basics; Kubernetes/OpenShift fundamentals; logging/monitoring patterns; secrets management and environment separation (DEV/UAT/PROD).

Main Tasks and Responsibilities

1. Build Python Services and Agentic Components
• Develop production-grade backend services and worker processes aligned to the defined solution architecture.
• Implement orchestration components: job queues, scheduling, state management/state machines, retries, idempotency, and compensating actions.
• Build and maintain tool connectors/integrations with enterprise systems (APIs, databases, files), following safe execution boundaries and permission controls.
• Contribute reusable libraries and shared components to accelerate delivery across multiple client solutions.
2. Implement Reliable LLM and RAG Capabilities
• Integrate LLM capabilities into services using tool/function calling, structured outputs, and strict schema validation.
• Develop and maintain RAG pipelines: ingestion, indexing, retrieval, grounding, and evidence capture/citations where required.
• Apply context engineering practices: selecting/structuring context, minimizing irrelevant context, maintaining traceability, and improving response determinism.
• Implement guardrails and safety controls: input validation/sanitization, output validation, refusal/fallback handling, and policy-aligned tool usage.
3. Testing, Quality, and Release Discipline
• Build and maintain test suites: unit, integration, and regression testing (including prompt/agent regression where applicable).
• Participate in code reviews and follow engineering standards for maintainability, security, and correctness.
• Use GitHub-based workflows effectively: PR hygiene, branching strategies, code owner reviews, and CI/CD integration.
• Support release processes with strong documentation, configuration discipline, and readiness checks.
4. Observability, Performance, and Operational Readiness
• Implement logging, tracing, metrics, and decision logs for services and agent runs; support run replay and incident investigation.
• Profile performance bottlenecks and optimize latency, throughput, and cost across critical paths.
• Contribute to dashboards, alerts, runbooks, and operational procedures to maintain stable production systems.
5. Security, Compliance, and Enterprise Delivery
• Implement secure coding practices, secrets handling, and least-privilege patterns in tool execution and integrations.
• Follow enterprise governance expectations: audit-friendly logs, change controls, environment separation, and controlled deployments.
• Collaborate closely with the Agentic AI Architect, infra Teams, COE to deliver compliant, production-ready solutions.