Job Description: About the Role
EXL is looking for an Applied Agentic AI Architect to design, build, and productionize autonomous and semi-autonomous AI agent systems that power real business outcomes for our clients. This is a hands-on architecture role: you will own the technical design of multi-agent and conversational AI solutions end to end — from orchestration patterns and model selection to evaluation, guardrails, and deployment at scale.
You will sit at the intersection of applied research and engineering, translating fast-moving agentic AI capabilities into robust, secure, and cost-effective solutions across domains such as insurance, healthcare, banking, and customer operations. You will work directly with client stakeholders, data scientists, and engineering teams to move solutions from proof-of-concept to enterprise production.
Key Responsibilities
- Architect agentic systems. Design multi-agent architectures — planner/executor patterns, tool-calling, memory, retrieval, and human-in-the-loop workflows — that are reliable, observable, and maintainable in production.
- Build with modern agent frameworks. Implement solutions using frameworks such as LangGraph and Google's Agent Development Kit (ADK) , selecting the right orchestration pattern for each use case.
- Design conversational AI. Architect enterprise conversational and voice AI experiences — including contact-center and customer-experience solutions on platforms such as GECX — covering intent handling, dialog management, escalation, and containment.
- Model strategy & selection. Evaluate and integrate frontier LLMs including Gemini , Grok , and others; make principled trade-offs across accuracy, latency, context window, cost, and deployment constraints.
- Retrieval & grounding. Design RAG pipelines, knowledge integration, and grounding strategies to reduce hallucination and improve factual reliability.
- Evaluation & guardrails. Establish evaluation harnesses, offline/online metrics, red-teaming, safety guardrails, and monitoring for agent behavior in production.
- Productionize & scale. Own MLOps/LLMOps concerns — CI/CD for prompts and agents, versioning, observability, caching, and cost optimization.
- Lead technically. Set architecture standards and reusable patterns, review designs, mentor engineers and data scientists, and act as a trusted technical advisor to clients.
- Stay ahead. Track the rapidly evolving agentic AI landscape and translate new techniques into practical, differentiated client offerings.
Required Qualifications
- Bachelor's or Master's in Computer Science, AI/ML, Engineering, or equivalent practical experience.
- 8+ years in software/AI engineering, with 3+ years building and shipping production ML/GenAI systems (architecture-level experience strongly preferred).