Maxxton
Website:
maxxton.com
Job details:
Position:- Lead - Agentic AI (Backend, Java)
Designation:- Lead Software Engineer
Experience:- 5-12 years
Location:- Remote / Hybrid - India
Who We Are Looking For
- We are looking for an Agentic AI Lead (Backend, Java) to architect and ship production-grade, agent-powered systems for our hospitality platform.
- You will lead the design and build of multi-agent backends that power guest-facing experiences (booking, concierge, in-stay support) and property and revenue operations (reservations, pricing, ops automation).
- Building agentic products is deeply cross-functional. Tight collaboration with platform, product, data, and hospitality-domain teams is essential, and a strong sense of ownership with a commitment to engineering excellence is required.
What You Will Need To Do
- Own the architecture and delivery of agentic AI backends in Java and Spring Boot — tool-calling agents, multi-agent orchestration, planner-executor patterns, memory, and evaluation harnesses.
- Integrate with foundation model providers (OpenAI, Anthropic, AWS Bedrock) and build provider-agnostic abstractions so models can be swapped on cost, latency, or capability.
- Design and implement Model Context Protocol (MCP) servers and tool wrappers that expose Maxxton's hospitality capabilities — reservations, PMS, revenue management, guest profiles — safely and reliably to agents.
- Engineer the agent runtime end-to-end: context and prompt engineering, tool selection, guardrails, rate limiting, cost controls, retries, idempotency, and tracing.
- Define and run evaluation strategies — offline eval suites, online A/B tests, regression catches, and human-in-the-loop review — so agents demonstrably improve in production.
- Lead and mentor a pod of backend engineers: run design reviews, set the coding bar, drive code reviews, and partner with product on the agentic roadmap.
- Collaborate with frontend, data, infra, and hospitality-domain experts to ship end-to-end agent experiences that delight guests and operators.
- Track the rapidly evolving agentic ecosystem (frameworks, models, MCP, multi-agent patterns) and translate it into pragmatic engineering choices for the team.
- Keep the ticketing system (JIRA) updated with estimations, progress, blockers, and due dates; raise risks early and clearly.
- Champion engineering best practices: secure-by-default design, automated testing, CI/CD, observability, and clean software architecture.
What You Will Need To Have
- Bachelor's or Master's degree in Computer Science, Engineering, or equivalent practical experience.
- 5–12 years building production backend systems in Java, with deep expertise in Spring Boot, Hibernate/JPA, REST and microservices.
- 1–2+ years hands-on building LLM-powered or agentic systems in production — tool/function calling, RAG, planners, or multi-agent orchestration.
- Strong working knowledge of foundation model APIs (OpenAI, Anthropic, AWS Bedrock) and the trade-offs between them across cost, latency, context, and capability.
- Hands-on experience with Model Context Protocol (MCP) and/or comparable tool-orchestration approaches (function calling, ReAct, planner-executor, agent graphs).
- Solid grounding in prompt and context engineering, structured outputs, evaluation, and observability for non-deterministic systems.
- Comfort with cloud platforms (AWS, GCP, or Azure), containers (Docker, Kubernetes), event-driven architectures, and message brokers (Kafka, RabbitMQ).
- Track record of technical leadership — owning design for non-trivial systems, mentoring engineers, and driving cross-team decisions.
- Hospitality, travel, or booking-platform experience is a strong plus; familiarity with PMS, channel managers, or revenue management is highly valued.
- Sound logical, analytical, and problem-solving skills.
- Excellent communication — able to influence engineers, product, and senior stakeholders with clarity.
We are looking to grow our teams with people who share our energy and enthusiasm for creating the best experience for our customers.
Click on Apply to know more.