Predicting Follow-On Revenue Across 4M Active Customers and 15+ Payments per Second
Sun King delivers off-grid solar energy to communities without reliable electricity. With customers in 46 countries worldwide, Sun King has sold over 27 million solar products and extended 1.2 billion in solar loans to customers since 2007. Most customers pay over time using a pay-as-you-go model, many without access to formal banking systems.
To keep expanding, Sun King needed sharper visibility into credit risk and revenue realization. Their ability to plan, price, and invest in infrastructure depends on accurate predictions built from vast streams of customer data.
The Challenge: Scaling Insights Across Millions of Customers
Sun King manages bursts of 91 payments per second. Each transaction adds to a growing, complex dataset. Their team needed a better way to process this information to:
- Predict follow-on revenue realization (FRR) at a granular level
- Simulate individual credit risk and price products accordingly
- Reduce delays caused by tool limitations and memory constraints
- Collaborate effectively across data science, finance, and operations teams
Sun King’s pre-existing tools weren’t scaling effectively: workflows frequently stalled and rerunning workflows added unnecessary friction to the insight generation process.
Why Sun King Chose Zerve To Address This Challenge
Sun King selected Zerve, a cloud-native data and AI development and deployment platform, to give them the structure and speed they needed. As a collaborative, modular coding environment, Zerve helped the team organize complex workflows, run data calls in parallel, and reuse experiments across projects while maintaining version control. Key improvements included:
- Data retrieval time slashed in half: Queries that previously took 24 minutes now complete in just 1 minute.
- Scalable processing: Parallelized calls handled 4.3 billion-row datasets without strain
- Cleaner collaboration: A shared canvas eliminated notebook sprawl
- Increased efficiency: Teams could run multiple experiments at once and repurpose components
Zerve made it easier to turn raw payment data into usable signals without constant tool maintenance.
The Impact: Faster Analysis, Smarter Forecasting
Zerve helped Sun King accelerate every part of their workflow. Analysts could go from data retrieval to model execution without running into memory constraints. Teams could test new risk models across 4 million active users in one environment.
The block-based architecture offered two major advantages: first, it allowed persistent storage of intermediate outputs, eliminating the need to restart notebooks; and second, it facilitated better code organization compared to maintaining multiple Jupyter notebooks for similar tasks.
"We were working with a 4.3 billion-row dataset, and a high-end notebook instance could only handle 18 million-rows at a time without going out of memory due to the nature of preprocessing to be done on the data. That meant running the same call 24 times in a row. With Zerve, we could run all 24 in parallel. That shift alone saved us a huge amount of time and let us move faster on model development”
- Arindam Bose - Head of Data Science
Sun King now relies on Zerve across its data science and finance operations. With reliable insights and a smoother workflow, their team is better equipped to respond to business needs and market shifts.
What’s Next: Expanded Use of Zerve
Encouraged by these results, Sun King is expanding their use of Zerve to support:
- More dynamic credit risk modeling
- Automated sales forecasting
- Operational reporting across new markets
Bottom Line
Zerve helped Sun King shift from reactive reporting to proactive planning. With clearer insights and better performance, Sun King is making faster, more confident decisions that directly impact revenue, risk, and growth.
Read more about Sun King and Zerve on AI Chief.
