Smart charging for a smarter grid
How I helped pioneer BMW's first dynamic charging system — using machine learning to balance power grid load while creating a delightful experience for electric vehicle drivers.
The challenge
As BMW launched their electric vehicle line, we discovered an unexpected issue. Most drivers plugged in between 6–8 PM, creating massive spikes in power demand that threatened to overwhelm grids and raise electricity costs for everyone.
We needed charging that intelligently distributed itself through the night while ensuring every driver had a full battery by morning. The real challenge: making grid optimization feel magical, not restrictive.
Peak load distribution — before smart charging
Research & discovery
Understanding drivers' needs, anxieties, and behaviors around EV charging.
Michael, 42Software engineer · Palo Alto · 65-mi commute
- Granular control over charging parameters
- API access for home automation
- Detailed analytics and usage data
“I need my car's charging to integrate seamlessly with my solar panels and time-of-use rates.”
Sarah, 38Sustainability director · Berkeley · 28-mi commute
- Carbon footprint tracking
- Green energy source preferences
- Community impact visibility
“Show me exactly how my charging choices affect the grid's renewable energy usage.”
David, 48Sales executive · San Jose · variable commute
- Guaranteed readiness for unexpected trips
- Cost optimization without complexity
- Set-and-forget reliability
“I need my car ready to go, but I don't want to think about it or pay more than necessary.”
Range anxiety is real.
73% of drivers worried their car wouldn't be charged when needed — even with smart scheduling. Guarantees had to come before optimization.
Transparency builds trust.
Users wanted to see exactly when and why their car would charge at specific times. Visibility was the feature.
Money talks.
Showing actual dollar savings was 3× more effective than environmental benefits alone.
The solution
An intelligent system that makes doing the right thing feel effortless.
Predictive algorithm
Co-developed an ML algorithm that predicted optimal charging times from grid load and individual driving patterns.
Transparent control
An interface giving drivers full visibility and control over their charging schedule — override anytime, no penalty mystery.
Incentive design
A rewards system that made grid-friendly charging feel like winning, not compliance.
Before / after — the charging flow
The shipped v1 charged the moment you plugged in. The redesign moved charging into the overnight off-peak window while guaranteeing readiness by departure time.
The impact
Key learnings
Make the right choice the easy choice.
By optimizing for user convenience first, grid optimization happened as a natural outcome.
Transparency builds trust.
Showing users exactly how the system worked made them far more likely to participate.
Small incentives drive big changes.
Gamification turned charging from a chore into something drivers looked forward to.
Design for the ecosystem.
Considering drivers, utilities, and BMW together produced a solution that worked for everyone.