ML Operations Engineer

Wingfin Singapore; Kuala Lumpur; Hybrid; MYR 120,000 - 180,000; Full time

Job summary

Wingfin Singapore is seeking an ML Operations Engineer to build and run the infrastructure turning driving data into trained models. You will oversee production ML platforms, data and HPC workloads across Malaysia or Singapore, ensuring fast, reliable and reproducible training pipelines.

You will manage in-house and cloud GPU capacity, data pipelines and large-scale storage, while maintaining rigorous versioning, logs and runbooks for smooth handovers and reproducibility.

Qualifications

• Bachelor's degree or higher in Computer Engineering, Computer Science or a related field.
• 3+ years building or operating production infrastructure for ML, data, trading or HPC workloads.
• Strong Python. Linux, shell and networking: SSH, reverse proxies, ports, TLS.
• Cloud GPU platforms (AWS, GCP, Azure) with object storage.
• GPU marketplaces like Vast.ai.
• Cloudflare R2 object storage, DNS and Tunnels.
• Docker and Kubernetes.
• Relational databases (PostgreSQL, MySQL).
• GPU environments: CUDA, drivers and multi-GPU machines.
• Use of modern AI tools for coding, debugging, operations.
• Ownership through to completion.
• Excellent attention to detail.

Responsibilities

• Training platform: Run and extend the platform to submit, schedule and track training jobs.
• GPU compute: Operate in-house and cloud GPU capacity, scheduling, access and cost reporting.
• Data pipelines: Turn raw vehicle recordings into validated, training-ready datasets.
• Storage: Manage large-scale data storage and transfer between sites and cloud.
• Deployment automation: Automate deployment and upgrades of services without interrupting jobs.
• Reproducibility: Track experiments, datasets and model artefacts for full traceability.
• Model packaging: Package trained models for simulation, testing and deployment.
• Observability: Monitoring, logging and alerting for jobs, machines and pipelines.
• Documentation: Runbooks and handovers for operations.

Skills

Python

Linux

Shell scripting

Networking

Cloud GPU platforms

GPU marketplaces

Docker

Kubernetes

PostgreSQL

MySQL

CUDA

AI tools

Ownership

Attention to detail

Education

Bachelor's degree or higher in Computer Engineering, Computer Science or related field

Tools

Cloudflare

R2 object storage

AWS

GCP

Azure

Job description

Wingfin develops autonomous driving systems. Our mission is rooted in accessibility: human-like autonomy for safe, cost-efficient, real-world scalability. By leveraging cutting-edge deep learning and end-to-end neural networks, Wingfin is redefining the landscape of self-driving technology. We are building a practical, cost-efficient platform that brings safe autonomy to every vehicle on the road. Headquartered in Singapore, the company boasts an international team of multi-talented experts. We are looking for talented and motivated individuals to join our team in Malaysia or Singapore.

The Role

We are looking for an ML Operations Engineer to build and run the infrastructure that turns our driving data into trained, tested models. You will have several years of hands‑on experience operating production ML, data or high‑performance computing platforms, and you will own ours end to end.

Our autonomous‑driving models are trained on large volumes of real and simulated driving data, using both in‑house and cloud GPUs. Your job is to keep training fast, reliable and reproducible: data in, jobs scheduled, models out, and every result traceable to its code, data and configuration.

What You Will Work On

• Training platform. Run and extend the platform the ML team uses to submit, schedule and track training jobs.
• GPU compute. Operate in‑house and cloud GPU capacity: scheduling, access and cost reporting
• Data pipelines. Turn raw vehicle recordings into validated, training‑ready datasets
• Storage. Manage large‑scale data storage and transfer between sites and cloud.
• Deployment automation. Automate deployment and upgrades of services without interrupting running jobs.
• Reproducibility. Track experiments, datasets and model artefacts so that any result can be rebuilt from its exact inputs.
• Model packaging. Package trained models for simulation, testing and deployment.
• Observability. Monitoring, logging and alerting for jobs, machines and pipelines.
• Documentation. Runbooks and handovers that let others operate the systems you build.

What We Are Looking For

• Bachelor's degree or higher in Computer Engineering, Computer Science or a related field.
• 3+ years building or operating production infrastructure for ML, data, trading or HPC workloads.
• Strong Python. Solid Linux, shell and networking: SSH, reverse proxies, ports, TLS.
• Cloud GPU platforms. Hands‑on experience running GPU workloads on AWS, GCP or Azure, including their object storage.
• GPU marketplaces. Running training workloads on Vast.ai or similar.
• Cloudflare. R2 object storage, DNS and Tunnels.
• Docker and Kubernetes.
• Relational databases (PostgreSQL, MySQL or similar).
• GPU environments: CUDA, drivers and multi‑GPU machines.
• Use of modern AI tools is a requirement. We expect active use of AI‑assisted tools for coding, debugging, operations and day‑to‑day engineering productivity.
• Ownership through to completion. You report status in writing, raise a blocker on the day it appears with the decision you need, and leave a usable handover before any absence.
• Excellent attention to detail. A silently changed dataset, a stale cache or a mismatched configuration can invalidate weeks of training.

Useful, But Not Required

• PyTorch training at scale
• Large video or image datasets
• Robotics or autonomous systems
• ONNX or TensorRT
• C++ and performance profiling

The Kind Of Person We Want

You might be a good fit if you have ever:

• Cut a pipeline's run time by profiling it rather than buying a bigger machine.
• Rebuilt a deployment from code after a machine failed, and it came back identical.
• Found that a training result was wrong because the data underneath it had changed.
• Shipped an infrastructure upgrade that nobody's running jobs noticed.
• Written the runbook you wished someone had left for you.

What We Offer

• Ownership of the training infrastructure for a real autonomous‑driving programme.
• Daily work with the ML team: your platform is used every day, and you hear at once what works.
• Large real‑world driving datasets and serious GPU compute.
• A technically demanding environment where good engineering matters more than titles.