Agentic Architect

Overview

Lead the architecture of agentic AI solutions for banking and financial services clients, from business problem through target architecture, integration and deployment blueprint.

What you'll do

• Lead solution design from business problem to target architecture, integration and deployment blueprint.
• Design single and multi-agent workflows, tool calling, memory, retrieval and human-in-the-loop controls.
• Define security, resilience, latency, cost, observability and scalability requirements.
• Select models, cloud services, vector search and frameworks; assess build versus buy.
• Set standards and guardrails, review designs, and partner with product, cloud and AI governance.
• Embed AIDLC in design: stage gates, governance checkpoints and architecture decision records.

What you'll need

• Architecture and distributed systems.
• Hands-on software engineering.
• Secure cloud architecture.
• API, data and identity integration.
• LLM, RAG and agentic patterns.
• Tool calling, state and memory.
• Agent evaluation.
• Build-versus-buy trade-offs.
• Least-privilege tool access.
• Prompt-injection defence.
• Approval gates for risky actions.
• Audit trails, model risk and residency.
• Defines AIDLC stages and gates.
• Evaluation and release criteria.
• Governance checkpoints.
• Architecture decision records.

Nice to have

• LangGraph, Semantic Kernel, AutoGen.
• Enterprise search and vector databases.
• Event-driven systems and microservices.
• Red-teaming and LLM observability.
• Model governance and AI safety.
• Disaster recovery and FinOps modelling.

Details

• Location: India.