ABOUT CLIENT
Our client is a big fintech company from Japan
JOB DESCRIPTION
Responsible for the design, construction, and upkeep of production-grade AI agent systems, including areas such as context engineering, instruction architecture, secure execution boundaries, tool integrations, multi-step orchestration, memory strategies, and reliability patterns.
Manage the complete lifecycle of agents from prototyping to deployment, monitoring, and iterating.
Develop and maintain an evaluation pipeline to measure agent quality, identify regressions, and enforce deployment gates through the use of golden datasets, scenario suites, and automated checks.
Instrument agents and agent platforms for production observability, such as structured logging, tracing, metrics, latency monitoring, cost monitoring, and analysis of tool-call success rates and failures.
Establish operational readiness standards, including rollback criteria, incident response playbooks, and recovery paths for common failure modes.
Collaborate with product engineering teams to identify high-value use cases suitable for agent automation, operating in a Central Agent Ops role to enable AI product builders through AI enablers.
Translate business workflows into tasks executable by agents and provide coaching to engineers on context engineering best practices, harness design, regression testing patterns, agent skill design, and tool-contract discipline.
Streamline the onboarding process for teams adopting AI capabilities and train product engineers to independently extend and maintain agent skills.
Develop and maintain organizational standards for agents, including naming conventions, context file structures, skill interface contracts, evaluation criteria, and release quality benchmarks.
Establish and enforce "repo-as-discipline" practices to ensure that agent knowledge is versioned, reviewable, discoverable, and reusable.
Cultivate a shared agent skills library for teams to reuse and extend, while keeping track of AI tooling/framework updates and external best practices to provide centralized information to product teams.
Facilitate internal knowledge-sharing sessions, showcases, and retrospectives to efficiently propagate learnings.
JOB REQUIREMENT
Extensive 12+ years of industry experience in building and deploying production AI agents using modern frameworks.
Proficient in context engineering, including instruction architecture, token management, caching strategies, and latency-aware design.
Experienced in developing evaluation pipelines, automated quality gates, and regression detection.
Familiar with agent observability, including tracing, structured logging, latency, and cost monitoring, as well as tool-call reliability metrics and failure analysis.
Capable in designing guardrails, output validation, prompt injection mitigation, and safe execution boundaries for tools/actions.
Strong backend engineering skills with the ability to own services/APIs end-to-end.
Effective communicator with the ability to coach engineers, facilitate cross-team discussions, and write clear technical documentation.
Experience in production reliability and platform operations, including event-driven architectures, retries/backoff, DLQs, idempotency, ordering, backpressure, CDC/outbox-style patterns, Kubernetes-based deployment and day-2 operations, CI/CD pipelines, infrastructure as code, on-call, incident response, postmortems, and SRE-style practices.
Experience with RAG systems, ingestion, chunking, embeddings, hybrid search, and retrieval evaluation.
Familiarity with MCP/Model Context Protocol or similar agent tooling standards and tool integration ecosystems.
Proficiency across Java/Kotlin (Spring Boot) and Python in production environments.
Engineers with SRE/DevOps background transitioning into AI, who naturally think about reliability, observability, and incident response.
Backend engineers with hands-on LLM/agent framework experience, willing to work cross-functionally and enable multiple teams.
MLOps/LLM engineers interested in embedding in product organizations and shipping applied systems, not only model infrastructure.
Engineers who prioritize documentation, standards, and knowledge transfer as first-class engineering outputs.
WHAT'S ON OFFER
CONTACT
We are
PEGASI – IT Recruitment Consultancy in Vietnam. If you are looking for new opportunity for your career path, kindly visit our website
www.pegasi.com.vn for your reference. Thank you!