Manager, GenAI L3 Support Engineer, Group Technology & Ops

UOB; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

United Overseas Bank Ltd. seeks a Manager, GenAI L3 Support Engineer to own production incident response for GenAI use cases, triage complex bugs, and coordinate rapid fixes.

You will work with LangChain, Elasticsearch, Redis, and guardrails within a Kubernetes/OpenShift environment to ensure reliability. The role emphasizes building robust pipelines, improving observability with OpenTelemetry and Grafana, and leading post-incident reviews with stakeholders across groups.

Qualifications

• 3 to 4 years of software engineering or SRE with AI/data in production.
• Strong Python and experience with LangChain projects.
• Hands-on with Elasticsearch and Redis in production environments.
• Experience in observability stacks (Prometheus/Grafana/OpenTelemetry).
• Familiarity with Kubernetes/OpenShift and CI/CD (Jenkins).

Responsibilities

• Own L3 incident response for GenAI use cases in production.
• Triages, debugs, fixes defects or coordinates vendor fixes.
• Root-cause complex bugs across LangChain flows, Elasticsearch and Redis.
• Build, test and deploy code/config updates through Jenkins with approvals and rollbacks.
• Harden pipelines with input validation, timeouts, retries and fallbacks.
• Improve RAG quality and tune latency for stability.
• Define golden signals, OpenTelemetry traces, and dashboards in Grafana.
• Maintain runbooks, playbooks and post-incident RCAs.
• Coordinate with platform/infra on OpenShift, GPUs, vLLM, and NVIDIA services.
• Ensure secure data handling and compliant logging practices.

Skills

Python

LangChain

FastAPI

Flask

Async patterns

Task queues

Git

Kubernetes

OpenShift

Elasticsearch

Redis

Prometheus

Grafana

OpenTelemetry

Tools

Kubernetes

OpenShift

LangChain tooling

Jenkins

Job description

## Manager, GenAI L3 Support Engineer, Group Technology & OpsApply: Onsite: North-East Region (City Area): Full time: Posted Yesterday: JR97155Company: 1011 United Overseas Bank Ltd **About UOB**United Overseas Bank Limited (UOB) is a leading bank in ASEAN with a global network in Southeast Asia, Asia Pacific, Europe and North America. Operating through our head office in Singapore and banking subsidiaries in China, Indonesia, Malaysia, Thailand and Vietnam, we have a global network of about 430 branches and offices in 19 markets. At the heart of UOB is our culture, shaped by the UOB Way and anchored on our four values – Honourable, Enterprising, United and Committed. For more than 90 years, these values have guided how we do right by our customers, collaborate with one another and create long-term value for the communities we operate in. As One Bank, we are committed to helping our colleagues build sustainable careers grounded in purpose, supported by strong values, and enriched with meaningful opportunities to grow. **Job Description*** Own L3 incident response for GenAI use cases in production. Triage, deep debug, fix or work with the vendor to resolve defects and improve reliability* Review, reproduce and root cause complex bugs across LangChain flows, Elasticsearch queries, Redis caches and safety layers such as Guardrails AI and LlamaGuard* Implement safe changes and hotfixes. Build, test and deploy code and config updates through Jenkins with proper approvals and rollbacks* Partner with business users to understand issue impact and edge cases. Translate vague problem reports into actionable steps and clear acceptance criteria* Harden use case pipelines. Add input validation, timeouts, retries, circuit breakers and fallback strategies to reduce customer facing impact* Improve RAG quality for stability. Tune chunking, retrieval parameters, query rewriting and caching to reduce latency and errors* Maintain and evolve runbooks, playbooks and knowledge articles. Keep them current and prove they work with regular game days* Drive observability for the use cases. Define golden signals, add OpenTelemetry traces, wire up Prometheus metrics and curate Grafana dashboards* Manage production hygiene. Track error budgets, SLOs and SLAs. Push for defect burn down and change quality* Coordinate with platform and infra teams when incidents touch OpenShift, GPUs, vLLM or NVIDIA Enterprise AI services* Perform safe data operations. Validate indices, manage Redis eviction strategies and handle backfills and reindex tasks with minimal risk* Champion secure by default practices. Enforce secrets handling, PII redaction and prompt or output guardrails* Lead post incident reviews with blameless RCAs and concrete corrective actions. Close the loop with business stakeholders* Proactively surface reliability risks and propose code or architecture changes that remove recurring failure modes Job Requirement* 3 to 4 years of software engineering or SRE with at least one year supporting AI or data intensive services in production* Strong Python. Comfortable reading and fixing LangChain code, FastAPI or Flask services, async patterns and task queues* Hands on with LangChain in real projects. Chains, tools, retrievers, memory and callback handlers* Elasticsearch proficiency. Query DSL, relevance tuning, index lifecycle, scaling, snapshots and Kibana for triage* Redis proficiency. Caching patterns, pub or sub, streams, eviction, persistence and troubleshooting latency or timeouts* Safety and guardrails experience. Guardrails AI configuration, schema and validator design, LlamaGuard or similar classifiers in the loo* CI or CD with Jenkins. Declarative pipelines, approvals, artifacts, environment promotion and rollback strategy* Observability depth. Prometheus metrics, Grafana dashboards, OpenTelemetry traces and logs.* Able to decide which signals matter and set actionable alerts* Solid Git. Branching, pull requests, code reviews and release tagging. Comfortable with feature flags and canary or blue green patterns* Working knowledge of Kubernetes and OpenShift as a strong plus. Debugging pods, logs, events and basic resource tuning* Familiarity with LLM serving stacks such as vLLM or NVIDIA Enterprise AI is a plus. Know how to read model server logs, timeouts and token throughput* Data handling discipline. Understanding of PII, masking, prompt redaction and safe logging* Clear communicator who can write crisp incident updates, RCAs and user facing notes