ML Infrastructure Engineer

ML Infrastructure Engineer
Menlo Park, CA | On-Site | Full-Time/Direct Hire

Looking for ML Infra experts (Bay Area preferred) with deep experience in CUDA, GPU optimization, VLLMs, and LLM inference—pure language focus, no vision/audio.

Client Opportunity | Through Phizenix

Phizenix, a certified minority and women-led recruiting firm, is hiring on behalf of an AI startup pioneering diffusion-based large language models—built for faster generation, multimodal integration, and scalable enterprise deployment.

We’re looking for a ML Infrastructure Engineer to help build the infrastructure that powers large-scale model training and real-time inference. You’ll collaborate with world-class researchers and engineers to design high-performance, distributed systems that bring advanced LLMs into production.

Responsibilities

• Design and manage distributed infrastructure for ML training at scale

• Optimize model serving systems for low-latency inference

• Build automated pipelines for data processing, model training, and deployment

• Implement observability tools to monitor performance in production

• Maximize resource utilization across GPU clusters and cloud environments

• Translate research requirements into robust, scalable system designs

Must-Haves

• Masters or PhD in Computer Science, Engineering, or a related field (or equivalent experience)

• Strong foundation in software engineering, systems design, and distributed systems

• Experience with cloud platforms (AWS, GCP, or Azure)

• Proficient in Python and at least one systems-level language (C++/Rust/Go)

• Hands-on experience with Docker, Kubernetes, and CI/CD workflows

• Familiarity with ML frameworks like PyTorch or TensorFlow from a systems perspective

• Understanding of GPU programming and high-performance infrastructure

Nice-to-Haves

• Experience with large-scale ML training clusters and GPU orchestration

• Knowledge of LLM-serving tools (vLLM, TensorRT, ONNX Runtime)

• Experience with distributed training strategies (e.g., data/model/pipeline parallelism)

• Familiarity with orchestration tools like Kubeflow or Airflow

• Background in performance tuning, system profiling, and MLOps best practices

At Phizenix, we’re committed to supporting diverse and inclusive teams. This is your chance to shape the systems that power the next generation of AI innovation. Let’s build the future—together.