Magnefy

Details

Positions: Machine Learning Intern — Partial Discharge Classification
Location: Palo Alto, CA / Hybrid
Innovation Area: Built Environment and Infrastructure; Grid Integration and Resiliency; Software and Data Analytics
Position Type: TomKat Center-supported
Compensation: $8,500
Internship Term: 8 weeks 

Apply Here

Magnefy is a Stanford-born deep-tech startup redefining predictive maintenance for critical power infrastructure-starting with transformers. We combine AI-powered, high-speed magnetic sensing with physics-informed digital twins to detect failures earlier and more accurately than traditional approaches. Our platform translates complex, multimodal operational data into actionable insights for grid operators and mission-critical facilities such as data centers. The core technology originated from Stanford research and is being validated in real-world deployments. Magnefy is backed by leading climate and deep-tech investors and supported by competitive federal grants and early customer traction. We are a small, execution-driven team building technology with real-world impact.

Responsibilities

Depending on your interests and the team's priorities, your project may include:

  • Option A: Baseline Classification Model
    • Work with labeled PD waveform datasets
    • Extract features from time-series signals (statistical, frequency-domain)
    • Train and evaluate classification models (Random Forest, SVM, simple neural networks)
    • Document model performance and insights
  • Option B: Feature Engineering Study
    • Explore different feature extraction approaches (FFT, wavelets, pulse shape metrics)
    • Compare which features best distinguish PD types
    • Visualize feature distributions across classes
    • Recommend feature set for production model
  • Option C: Data Pipeline & Labeling Tool
    • Help build tools for organizing and labeling waveform data
    • Create visualizations for domain experts to review and label samples
    • Document data quality and labeling guidelines
    • Support the feedback loop for model improvement
  • What You'll Learn
    • Applying ML to real sensor/signal data (not just clean datasets)
    • Feature extraction from time-series waveforms
    • Working with imbalanced and noisy real-world data
    • ML experiment tracking and documentation
    • Collaboration with domain experts
    • IoT and edge computing concepts

Who should apply

We’re looking for a curious, collaborative, and execution-driven Machine Learning Intern who’s excited to work at the intersection of AI, edge IoT, and real-world power systems. You enjoy getting your hands dirty with messy, high-frequency sensor data, learning quickly, and turning ideas into working models.

You thrive in small teams, take ownership, and bring a can-do mindset to open-ended problems. You’re comfortable experimenting, asking thoughtful questions, and iterating fast—whether that means trying a new model, digging into raw waveforms, or working closely with hardware and systems engineers.

If you’re innovative, resourceful, and motivated by impact—and want to see your ML work directly influence a product improving grid resiliency—we’d love to hear from you.

Required expertise

  • Coursework in machine learning or data science
  • Programming experience in Python
  • Familiarity with ML basics (classification, train/test splits, evaluation metrics)
  • Curiosity and willingness to learn new domains
     
Helpful But Not Required
  • Experience with pandas, NumPy, scikit-learn
  • Exposure to deep learning (PyTorch or TensorFlow)
  • Signal processing coursework or experience
  • Familiarity with time-series data
  • Personal projects involving ML
     
What You'll Deliver

By the end of your internship, you will have:

  • A working prototype or proof-of-concept for your project
  • Documentation of your approach, experiments, and findings
  • Presentation to the team summarizing your recommendation
  • Code contributed to our ML repository
Tools You'll Use
  • Languages: Python
  • ML: scikit-learn, PyTorch (as needed)
  • Data: pandas, NumPy, matplotlib
  • Collaboration: Git, Jupyter notebooks
  • Cloud: AWS (basic exposure)

Bonus: If you have a GitHub or portfolio, we'd love to see it!

Preferred skills/Majors

Currently pursuing BS/MS in Computer Science, Electrical Engineering, Data Science, or related field