AI Infrastructure Engineer

About the Role

Pokee AI is seeking an AI Infrastructure Engineer to design and enhance the backend systems supporting reinforcement learning (RL) trained AI agents. This position emphasizes building scalable training environments, high-efficiency inference serving solutions, and robust production infrastructure intended for enterprise applications. Remote work is supported, with preference for candidates located in Singapore or the US.

Key Responsibilities

• Design and improve scalable training and inference systems tailored to RL-based AI models.
• Enhance model deployment focusing on latency, throughput, and cost efficiency across cloud and edge devices.
• Create and manage continuous integration/continuous deployment (CI/CD) pipelines alongside experiment tracking and model versioning frameworks.
• Develop data pipelines encompassing collection, preprocessing, and computation of reward signals for training.
• Collaborate closely with research scientists to bring novel algorithms and model structures into production.
• Guarantee infrastructure adheres to enterprise standards concerning reliability, security, and compliance.
• Support deployments across cloud platforms, on-premise environments, and edge devices.

Experience and Skills Required

• At least three years’ experience in machine learning infrastructure, platform engineering, or related systems roles.
• Proficiency in Python and low-level programming languages such as Rust, C++, or Go.
• Hands-on experience with ML serving frameworks including vLLM, TensorRT, Triton, and ONNX Runtime.
• Experience managing container orchestration with Kubernetes and Docker, and familiarity with cloud services like AWS and GCP.
• Strong understanding of GPU computing, distributed systems, and performance profiling techniques.
• Knowledge of machine learning experiment tracking and pipeline orchestration tools such as MLflow, Weights & Biases, or Apache Airflow.

Additional Qualifications (Preferred)

• Expertise in on-device or edge inference optimization strategies such as GGUF quantization, TensorRT-LLM, CoreML, or QNN.
• Experience with on-premise GPU hardware setups, including NVIDIA DGX, Dell PowerEdge servers, or Lenovo ThinkStation.
• Background in supporting reinforcement learning training loops or real-time learning systems in production environments.
• Understanding of enterprise software security and compliance standards like SOC 2 and data residency considerations.
• Active contributions to open-source projects related to machine learning infrastructure.

Additional Information

• Salary range: USD 130,000 to 150,000 annually.
• Location: Remote, with preference for Singapore or US candidates.
• Employment type: Full-time.