Senior Engineer, Applied AI

EXL (NASDAQ: EXLS) is a global data and artificial intelligence ("AI") company that offers services and solutions to reinvent client business models, drive better outcomes and unlock growth with speed. EXL harnesses the power of data, AI, and deep industry knowledge to transform businesses, including the world’s leading corporations in industries including insurance, healthcare, banking and financial services, media and retail, among others. EXL was founded in 1999 with the core values of innovation, collaboration, excellence, integrity and respect.

We are headquartered in New York and have more than 60,000 employees spanning six continents. For more information, visit www.exlservice.com.

Role Title: Senior Engineer, Applied AI 
BU/Segment: Digital, AI Innovation and R&D
Location: Dublin, Republic of Ireland (Flexible hybrid working) 
Employment Type: Permanent
 

Summary of the role:

We are seeking a hands-on Senior Engineer in EXL’s AI Innovation and R&D team to build and productionize Generative AI and Agentic AI capabilities for client-facing initiatives and internal accelerators. This is an AI-first engineering role focused on agentic workflows, advanced retrieval systems, evaluation, and production readiness, with strong software engineering fundamentals to ship reliable features. You will collaborate closely with data science, backend engineering, and delivery teams to implement scalable, maintainable AI systems and contribute to EXL’s rapid innovation cadence.

As part of your duties, you will be responsible for:

AI Agent Engineering
•    Build and own production-grade agent systems end to end, from design through deployment and ongoing operation, including steady-state maintenance, reliability improvements, and operational support. 
•    Implement stateful, durable agentic workflows with clear checkpoints, safe retries, and human-in-the-loop steps for high-impact actions. 
•    Design agent architectures with planning, tool use, memory, and escalation, and proactively address common failure modes such as hallucinations, tool misuse, state errors, and loops. 
•    Build secure tool integrations using MCP-based connectors and/or tool registries to expose internal services and approved external SaaS APIs to agents. 
•    Implement advanced retrieval and grounding, including hybrid retrieval (vector plus structured), reranking, relevance tuning, and robust context assembly for grounded responses. 
•    Treat evaluation as an engineering discipline by creating offline datasets, regression gates, and online monitoring, and defining measurable success metrics such as task success, groundedness, and tool-call correctness that can gate releases. 
•    Instrument AgentOps and LLMOps by tracing full trajectories (retrieval, model calls, tool calls, outputs), attributing cost and latency per run, and alerting on drift and failure patterns using standard observability practices and tools. 
•    Develop self-improving and self-evolving agent loops using evaluator-driven optimization (generate, evaluate, refine) and/or RL-style approaches where rewards are verifiable (tests, constraints, correctness checks). 
•    Use agentic coding workflows beyond autocomplete, such as Cursor, Replit, Claude Code, Codex to accelerate delivery while maintaining engineering standards, and standardize repo guidance via AGENTS.md and CLAUDE.md-style instruction files.
•    Stay current with fast-moving GenAI and Agentic AI tooling and research, and translate relevant innovations into pragmatic implementations

Backend Engineering
•    Optimize systems for cost, latency, and quality using pragmatic routing (planner versus executor), caching, and model/provider selection across hosted and local models. 
•    Build robust backend services that power agents, including Python services and production APIs (for example FastAPI), Postgres, and event-driven or job execution patterns (for example Kafka, Airflow, workflow engines), with strong integration hygiene (OAuth, webhooks, rate limits, idempotency). 
•    Deploy and operate in real environments using Docker and Kubernetes, CI/CD, and infrastructure best practices, and contribute to runbooks and operational readiness as part of shipping.

Delivery and collaboration
•    Partner with cross-functional teams to translate problem statements into implementable AI solutions, and deliver components that fit broader architectures.
•    Contribute to engineering best practices, documentation, and reusable components that accelerate delivery across projects.
•    Participate in a culture of applied innovation where strong ideas are not only shipped, but als