A career in IBM Consulting is rooted by long-term relationships and close collaboration with clients across the globe. You'll work with visionaries across multiple industries to improve the hybrid cloud and AI journey for the most innovative and valuable companies in the world. Your ability to accelerate impact and make meaningful change for your clients is enabled by our strategic partner ecosystem and our robust technology platforms across the IBM portfolio
Own the design and hands-on build of agentic AI solutions on Microsoft Azure for a tier-one Financial Services client. You’ll translate business use-cases into secure, scalable agent workflows (RAG, tools/functions, orchestration), stand up the Azure landing pattern and LLMOps foundations, and guide squads to production with the right controls for regulated environments (data, security, MRM, audit).
What you’ll do (outcomes & responsibilities)
Solution & architecture: Design end-to-end Azure reference patterns for agentic systems (tool-using agents, planners, RAG), covering data, orchestration, security, and observability.
Build & enable: Lead hands-on implementation (PoCs, MVP, scale), pairing with engineers; codify patterns as reusable modules/accelerators.
LLMOps foundations: Set up Prompt Flow/evals, model registry, safety/guardrails, telemetry, rollback, and promotion workflows across environments.
Data for AI: Define/document retrieval patterns (chunking, embeddings, vector indices), grounding strategy, and latency/SLA budgets.
Security & compliance (FS): Implement least-privilege, private networking, Key Vault, managed identities, threat protection; align with DORA/FCA/PRA controls and Model Risk Management (MRM) practices.
Quality & safety: Establish eval suites (functional, factuality, toxicity, PII), red-team plans, and A/B or shadow testing for releases.
Delivery leadership: Break work into stories, estimate, and steer squads; collaborate with client architects, risk, and data owners.
Stakeholdering: Translate technical trade-offs into clear options; run design reviews and readiness gates to production.
Knowledge transfer: Document patterns, run playbacks, and mentor client and IBM teams.
Technical
5–8+ years building cloud solutions, with 3+ years on Azure and recent GenAI delivery to production.
Designed and shipped RAG and/or tool-using agent solutions (function calling, tool orchestration, multi-step workflows).
Practical with the Azure AI stack:
Azure OpenAI (prompting, function calling, Assistants/agents concepts), Azure AI Search (vector & hybrid), Prompt Flow/evaluations, Azure ML or AI Studio/Foundry, App Insights/Azure Monitor.
Data plane: ADLS Gen2, Delta/Parquet, Fabric or Databricks (nice if both), SQL/Cosmos/PosgreSQL with pgvector or vector stores, cache (Redis).
App/infra: Functions, Container Apps/AKS, API Management, Event Grid/Service Bus, Key Vault, Managed Identity, Private Link, VNets.
LLMOps/MLOps working knowledge: experiment tracking, evals, canary/shadow deploys, safety filters/guardrails, model/version governance.
Security-by-design on Azure: RBAC/PIM, network isolation, secrets/keys, data classification, audit & logging.
IaC & DevOps: Terraform or Bicep; GitHub (Actions, Packages, Advanced Security) or Azure DevOps; trunk-based or similar.
Performance & cost engineering for LLM apps (token/latency budgets, caching, chunking/embedding strategy).
Domain & delivery
Built solutions in regulated FS (banks/insurance/markets) with an understanding of DORA-style controls, data residency, MRM, and audit traceability.
Comfortable leading technical workstreams, story mapping, estimation, and guiding 4-8 engineers.
Behaviours
Clear, concise communicator with senior stakeholders.
Pragmatic, outcome-oriented; documents decisions and trade-offs.
Coaches others; leaves reusable patterns behind.
Agent frameworks: Semantic Kernel, AutoGen, LangChain; planning/tool-use strategies; function/tool catalogs.
Retrieval expertise: advanced chunking, hybrid search, re-ranking, query expansion, doc QA evals.
Safety & compliance tooling: Content filters, jailbreak detection, PII redaction; integration with Purview (data lineage/classification) and Defender controls.
Observability: custom telemetry for LLM calls, traces, user/session analytics, prompt/version lineage.
Fabric & Databricks side-by-side experience; Unity Catalog/governance patterns.
Event-driven architectures and workflow engines (Durable Functions, Dapr, Logic Apps).
Certs: AZ-305, DP-203, AI-102, AZ-400 (or GitHub Foundations/Actions); any financial-sector security accreditation.
Prior consulting at a major SI or vendor; pre-sales artefacts (estimates, SoWs, roadmaps).
Tools & StackAzure OpenAI • Azure AI Search (vector) • Prompt Flow/Evals • Azure ML/AI Studio (Foundry) • ADLS Gen2 • Fabric/Databricks • Cosmos DB / Azure SQL / PostgreSQL (pgvector) • Redis • Functions • Container Apps/AKS • API Management • Event Grid/Service Bus • App Insights/Azure Monitor • Key Vault • Managed Identity • Private Link • VNets • Terraform/Bicep • GitHub Actions