I had a chat with Troy, DS manager @ Kevala.

A few things I got out of the chat:

On Kevala

• key mission: make energy data more useful and drive decisions in energy policy and utility decisions
• customers: They work a lot with utilities as their main stakeholders, as well as some state and federal regulatory bodies
• offerings:
  • They have a user platform that shows granular data on the electric grid, including high voltage transmission lines and lower voltage distribution lines
  • A key data asset is their granular data on the lower voltage distribution lines that feed customers and EV chargers
• example works:
  • One of their data science teams models different energy resources like rooftop solar, batteries, EVs, and EV charging infrastructure
  • They help utilities plan their grids and identify where new energy needs will arise at a granular level
  • They can help utilities understand impacts on revenue and customer bills from technology adoption

resources

• News sources
  • Canary Media | Covering the clean energy transition
  • Utility and Energy Transmission & Distribution News | Utility Dive
• Course
  • Electric Utilities Fundamentals and Future Course
• Communities
  • Young Professionals in Energy- SF Bay Area
• Datasets
  • Grid Status API – ex: forecast electricity price in Texas
  • Energy Information Administration (EIA)
• Book
  • The Grid: The Fraying Wires Between Americans and Our Energy Future by Gretchen Bakke | Goodreads

DS projects in energy

• Forecasting:
  • Forecasting energy demand
  • Forecasting electricity prices
• Optimization:
  • Controlling/optimizing different energy assets like batteries, EV chargers
• Other Applications:
  • Image recognition (e.g. analyzing satellite imagery)
  • LLMs (e.g. parsing regulations/zoning PDFs)
  • Graph analysis (representing utility grid network as a graph)

What he looks for when hiring

• clean, documented, production-quality code that can be collaborated
• Familiarity with version control, cloud platforms (AWS, GCP)
• Already thinking about evaluation metrics and how to assess model performance
• Able to discuss next steps - other modeling techniques, deployment considerations
• Thinking holistically about the entire solution pipeline beyond just modeling on laptop, i.e. big-picture thinking about full lifecycle of a data science solution

projects he would work on

• EV Charging Analysis
  • Analyze data on locations of EV charging stations
  • Look at characteristics of neighborhoods that lead to more charger deployments
  • Examine speed of charger rollout in different areas
• Transportation/Mobility Analysis
  • Use datasets on people's travel patterns and destinations
  • Analyze bike networks and most used routes/locations
• Battery/EV Charger Optimization Modeling
  • Model integrated battery and EV charger systems
  • Optimize battery usage alongside EV charging patterns
  • Factor in pricing signals, grid carbon intensity by time of day
• New York City Subway/Transit Analysis
  • Mentioned NYC has open data on subway schedules that could enable analysis

what keeps him up at night

• The massive scale of EV charger deployment needed to support transportation electrification goals
• He cites a target of 1.2 million chargers needed in California by 2035/2040, but currently only around 100,000 public ports
• Acknowledges utilities were not designed for this magnitude of new electric loads
• But he finds it exciting that utilities are open to changing their processes to accommodate this transition

advice for a master's student

• Make the most of being in the Bay Area for networking and events outside just your masters program
• Attend energy/sustainability groups and events to learn from other professionals
• This exposure helped give him a better understanding of why energy is used/deployed in certain ways