Root Access · 1 day ago
Machine Learning Engineer -- Intern
Root-access is seeking a Machine Learning Engineer Intern to join their paid Summer 2026 internship cohort. The intern will contribute to various ML projects including dataset preparation, model experimentation, and benchmarking while helping to build AI-native developer tools.
Information Technology & Services
Responsibilities
Train, evaluate, and debug machine learning models (e.g., deep learning, classical ML, multimodal models) using Python, PyTorch, and related frameworks
Use our internal AI-powered tooling to accelerate model development, dataset preparation, experiment tracking, and deployment workflows
Help test features like dataset validation, automated hyperparameter search, model introspection, and inference/runtime integrations
Provide structured feedback on usability, model behavior, edge cases, and failure modes (you’re part of the product loop)
Build demo models, evaluation scripts, or experiment workflows that help us validate reliability and usability of the platform
Read academic papers, model cards, and technical documentation to cross-verify model performance and expected behavior
Qualification
Required
Hands-on experience training ML models (vision, NLP, or embedded/edge ML all welcome)
Knowledge of core ML concepts: model architectures, loss functions, optimization, evaluation metrics
Experience with ML toolchains and workflows (e.g., PyTorch Lightning, Hugging Face, ONNX, TensorRT, Weights & Biases)
Curiosity about how AI development tools could be radically better—and a desire to help shape that future
Master's in Mathematics, Data Science, or Engineering
Prior work or internship experience with model training, ML research, or applied AI engineering
Preferred
Hungry to contribute to an ambitious startup, with opportunities to go full-time
Benefits
Paid Summer 2026 internship
Company
Root Access
Helping teams configure and program mission-critical hardware with AI.
Funding
Current Stage
Early StageCompany data provided by crunchbase