AI Infrastructure Engineer

About the role

As an AI Infrastructure Engineer, you will build and optimize the systems that power Pokee’s RL-trained AI agents—from scalable training pipelines to high-performance inference serving across cloud and on-device deployments. You’ll ensure that our research breakthroughs translate into production infrastructure that enterprises can rely on.

Responsibilities

• Design, build, and maintain scalable training and inference infrastructure for RL-based AI agent models
• Optimize model serving for latency, throughput, and cost across cloud (AWS, GCP) and on-premise/on-device environments
• Develop and manage CI/CD pipelines, experiment tracking, and model versioning systems
• Implement efficient data pipelines for training data collection, preprocessing, and reward signal computation
• Collaborate with research scientists to productionize new algorithms and model architectures
• Ensure infrastructure meets enterprise requirements for reliability, security, and compliance (SOC 2, data residency)

Requirements

• 3+ years of experience in ML infrastructure, ML platform engineering, or a related systems role
• Strong proficiency in Python and systems-level languages (Rust, C++, or Go)
• Hands-on experience with ML serving frameworks (vLLM, TensorRT, Triton, ONNX Runtime, or similar)
• Experience with container orchestration (Kubernetes, Docker) and cloud infrastructure (AWS or GCP)
• Solid understanding of GPU computing, distributed systems, and performance profiling
• Familiarity with ML experiment tracking and pipeline orchestration tools (MLflow, Weights & Biases, Airflow, or similar)
• Experience with on-device / edge inference optimization (GGUF quantization, TensorRT-LLM, CoreML, QNN)
• Familiarity with on-premise GPU deployments (NVIDIA DGX, Dell PowerEdge, Lenovo ThinkStation)
• Experience supporting RL training loops or online learning systems in production
• Background in enterprise software with knowledge of security and compliance frameworks
• Contributions to open-source ML infrastructure projects

You want to join a small, elite team solving one of the hardest problems in AI—building agents that actually work in the real world. You’ll have direct impact on the product, access to cutting-edge research, and the opportunity to shape the future of enterprise AI from the ground up.