ML Solution Architect Intern

## The role We're looking for an ML Solutions Architect (Early Career) to join the team behind Nebius Token Factory's serverless inference and fine-tuning platform for open-source LLMs. Working alongside senior Solutions Architects, you'll take on real technical work, including building and testing LLM-based solutions, benchmarking, and inference optimization. This is a hands-on learning role with close mentorship from senior SAs. Strong performers will be considered for a full-time Solutions Architect position at the end of the program. ## Your responsibilities - Help build and test LLM-based solutions and applications using Token Factory's inference services, including multimodal models (text, vision, audio). - Assist senior SAs with prompt engineering, model selection, benchmarking, and inference optimization. - Run performance and quality experiments to support proof-of-concept work. - Contribute to internal tooling and automation that improves how the SA team delivers. ## Must-haves - Currently pursuing or recently completed a BSc/MSc/PhD in Computer Science, Machine Learning, or a related field. - Strong Python programming skills. - Hands-on generative AI experience, including with common ML frameworks (e.g., PyTorch, Transformers). - Strong communication skills, with a willingness to explain technical concepts to diverse audiences. ## Nice-to-haves - Experience deploying/serving LLMs with vLLM, SGLang, or TensorRT-LLM. - Familiarity with inference optimization techniques such as quantization, batching, caching, and routing. - Knowledge of model architectures and fine-tuning approaches. - Contributions to open-source ML/AI projects. ## Preferred technical stack - Programming Languages: Python. - ML Frameworks and Libraries: vLLM, SGLang, TensorRT-LLM, Transformers, OpenAI/Anthropic SDKs. - Frameworks for Agentic Pipelines: Langchain / Langsmith / smolagents / equivalent. - API and Web Frameworks: FastAPI, Flask. - MLOps and DevOps tools: Kubernetes (K8s), Docker, Git. - Cloud Platforms: AWS (SageMaker, Bedrock), GCP (Vertex AI), Azure (Azure ML).