Lead Forward Deployed Engineer

Sea; Singapore; On-site; SGD 160,000 - 220,000; Full time

Job summary

SeaCloud helps customers build and run AI workloads via serverless model APIs, GPU VMs, bare metal GPU infra, and enterprise AI apps. The role owns complex customer engagements, guiding discovery, scoping delivery plans, and coordinating SeaCloud and customer teams through production adoption.

You will lead architecture reviews, onboarding, and performance validation for deployments across managed and customer-managed environments, including Kubernetes-based patterns and bare metal handoffs.

Qualifications

• 6+ years engineering experience with customer-facing production systems.
• Proven ability to turn ambiguous requirements into delivered solutions.
• Strong Linux, networking, containers, Kubernetes, storage, observability skills.
• Proficient in Python and shell scripting for automation.
• Experience with GPU-backed AI workloads and model deployment.
• Ability to mentor engineers and lead cross-team execution.
• Excellent English communication and willingness to travel.

Responsibilities

• Own complex customer engagements as the technical owner.
• Lead discovery discussions to uncover objectives, workloads, performance, security and budgets.
• Translate requirements into a scoped delivery plan with milestones and responsibilities.
• Review assumptions and commitments to identify readiness gaps before implementation.
• Lead architecture reviews, workshops, progress discussions and incident communication.
• Drive onboarding and integration of enterprise AI apps into IT environments.

Skills

Linux

Networking

Containers

Kubernetes

Storage

Observability

Distributed systems

Python

Shell scripting

GPU workloads

Benchmarking

Multi-GPU

English communication

Tools

NVIDIA GPU diagnostics

NCCL tests

InfiniBand

RoCE

RDMA

Slurm

Ansible

PyTorch

TensorRT-LLM

Job description

About the Team

SeaCloud helps customers build and run AI workloads through serverless model APIs, dedicated model hosting, GPU virtual machines, and bare metal GPU infrastructure, and adopt enterprise AI applications for collaboration and workflow automation.

Job Description

Own Complex Customer Engagements

• Act as the primary technical owner for assigned strategic or technically complex customers.
• Lead discovery discussions that uncover business objectives, workload characteristics, performance targets, security needs, budgets, and operational constraints.
• Work from the customer’s perspective, identify the underlying problem, and explain feasible options and their trade-offs.
• Translate requirements into a scoped delivery plan with milestones, dependencies, acceptance criteria, and clear responsibilities across the customer and SeaCloud teams.
• Review pre-sales assumptions and technical commitments, identifying readiness gaps before implementation begins.
• Lead architecture reviews, technical workshops, progress discussions, and customer-facing incident communication.

Deliver Across Managed and Customer-Managed Environments

• Drive customer onboarding and integration of enterprise AI applications into existing IT environments, including identity and access management, enterprise systems, data sources, and business workflows. Partner with customers and internal engineering teams to resolve implementation challenges and support the transition from proof of concept to production adoption.
• Guide customers toward an appropriate combination of serverless APIs, dedicated hosting, GPU VMs, bare metal, networking, and storage.
• Lead API migration and model evaluation, validating quality, feature compatibility, reliability, and cost using representative customer workloads.
• Deploy and optimize customer-managed inference stacks, including model loading, GPU memory allocation, concurrency, context length, caching, and multi-GPU execution.
• Design or review Kubernetes-based deployment patterns and, where explicitly included in the engagement, help establish customer-managed orchestration and observability.
• For bare metal engagements, coordinate infrastructure handover and validate the software environment, connectivity, storage, and operational readiness against agreed specifications.
• Configure or guide the use of Slurm when required by the customer workload and delivery scope.

Lead Performance Validation and Acceptance

• Build reproducible evaluation and benchmark methods using realistic request sizes, concurrency, datasets, and workload patterns.
• Assess quality, time to first token, inter-token latency, tail latency, throughput, and throughput achieved within the agreed service objectives.
• Develop GPU sizing and capacity recommendations that account for model memory, runtime overhead, workload variability, availability requirements, and utilization.
• Explain token-based, GPU-hour, VM-hour, and contract-based cost trade-offs using the applicable product terms.
• Define and review GPU cluster acceptance tests covering GPU health, topology, inter-node communication, and storage performance.
• Investigate complex failures across applications, orchestration, drivers, GPUs, networking, and storage, engaging specialists with clear diagnostic evidence.

Establish Reliable Delivery Practices

• Lead production-readiness reviews, customer acceptance, stabilization plans, and operational handover.
• Coordinate incident response and hardware-related escalations with Support, SRE, Infrastructure teams, and vendors, including recovery validation after repair.
• Define safe access, change, rollback, and data-handling practices within customer engagements.
• Build reusable playbooks, automation, reference architectures, sizing tools, and escalation procedures.
• Mentor other FDEs through technical reviews, paired troubleshooting, and customer-engagement coaching.
• Partner with Customer Success on technical health, adoption risks, and evidence of customer value.
• Prioritize recurring field problems and work with Product and Engineering on improvements that benefit multiple customers.

Requirements

• Typically 6+ years of relevant engineering experience, including substantial responsibility for customer-facing production systems or complex technical delivery.
• A record of taking ambiguous customer requirements through implementation, validation, and operational handover.
• Strong Linux, networking, containers, Kubernetes, storage, observability, and distributed-systems fundamentals.
• Strong Python and shell scripting skills, with experience building maintainable automation and reusable technical assets.
• Production experience with GPU-backed AI workloads and practical depth in inference deployment, performance analysis, or ML infrastructure.
• Experience with GPU memory sizing, workload benchmarking, runtime compatibility, and multi-GPU performance.
• Working knowledge of GPU cluster topology, collective communication, and the network and storage dependencies of distributed workloads.
• Ability to define credible acceptance criteria, interpret benchmark limitations, and separate verified findings from assumptions.
• Strong judgment around scope, technical feasibility, risk, customer commitments, and escalation.
• Excellent customer communication, including executive-level explanations, difficult conversations, and calm incident updates.
• Ability to mentor engineers and lead execution across teams without relying on reporting authority.
• Strong written and spoken English, with willingness to travel for customer workshops, acceptance, and critical launch phases.

Useful Additional Experience

• vLLM, SGLang, TensorRT-LLM, Triton, or comparable serving systems.
• Quantization, tensor or expert parallelism, and large-model serving.
• Slurm administration, Ansible-based cluster configuration, NCCL tests, InfiniBand, RoCE, or RDMA diagnostics.
• Distributed training, PyTorch DDP/FSDP, checkpoint recovery, or SFT/LoRA workflows in customer-managed environments.
• NVIDIA GPU diagnostics, hardware fault isolation, and vendor/RMA coordination.
• ARM64/x86 compatibility, multi-architecture images, and GPU software-stack validation.
• Building delivery standards or customer engineering practices in an early-stage cloud business.