Energy

Customer Story / AmpUp

AmpUp Ships EV Charging Features Safely Every Sprint with Cloudbees Feature Management

October 15, 2026

15 minutes to read

About AmpUp: AmpUp is a hardware-agnostic EV charging platform that helps businesses and property owners deploy and manage EV charging on the hardware of their choice.

Industry: EV Charging Software

Geography: North America

Summary: AmpUp, the hardware-agnostic EV charging platform behind more than 2,500 customers across North America , uses Cloudbees Feature Management to separate deployment from release. The team deploys every two-week sprint, rolls features out in controlled stages, and can switch off a bad feature in seconds, limiting the impact on drivers and charging-site host

Challenge: AmpUp needed to ship customer-facing features every sprint while limiting the damage a bad release could cause, across a production system drivers depend on and an engineering team spread across time zones.

Product: CloudBees Feature Management

Results:

  • Rolled back bad features in seconds instead of waiting five to ten minutes or more for a Kubernetes redeploy.
  • Switched off only the broken feature while the rest of the release kept running.
  • Capped early driver rollouts at 5% of users, minimizing the number of users a broken feature could impact.
  • Let whoever is online switch off a problem feature, so the owning engineer fixes it during normal hours instead of overnight.
  • Leveraged built-in audit and RBAC capabilities to track every code change and improve audit posture.

Since 2018, AmpUp has helped businesses and property owners deploy and manage EV charging on the hardware of their choice.

Supporting more than 130 charger makes and models, AmpUp powers charging networks for over 2,500 North American customers. AmpUp adopted Cloudbees Feature Management to put every major feature behind a flag, control which users and sites see each release, and turn off problem features instantly from any time zone, all without changing its existing delivery tools, including GitHub Actions.

Challenge

AmpUp has about 50+ employees, and its engineering team ships new features every two-week sprint. Most customers are in the US, and engineers work across time zones. Lam Nguyen, Senior Software Engineer, saw the risk with his very first assignment: a large, customer-facing change that needed a phased release. When asked about how he felt about delivering on this change, he said, "That's very risky for me, especially when I consider the greater impact that this feature change has on customers.” He knew it needed a feature flag and an incremental rollout.

AmpUp's release model created several pain points:

  • All-or-nothing rollbacks: eleases bundle many features, and they aren't all ready at the same time. Without flags, one bad feature means rolling back the whole release and redeploying the Kubernetes cluster.
  • Time-zone gaps: A bug found during US business hours meant either waking the engineer who built it or leaving customers with the problem for a full day.
  • One-size-fits-all releases: Drivers, businesses, and site hosts need new features to reach the right group first, not everyone at once.
  • Growing access, growing risk: 10+ engineers had flag access, and every flag change was a production change. AmpUp needed clear rules on who can change what, where, and with whose approval.

Solution

Nguyen had used Cloudbees Feature Management at a previous company and brought it to AmpUp. The team started on the free tier. Within 12 to 18 months, AmpUp signed a commercial agreement, flags were already how it released software.

AmpUp's rule is simple: if a tool isn’t the product it delivers to customers, it buys the tool instead of building it.

"We always find a tool first. We only consider building it when it isn't available in the market."

Lam Nguyen, Senior Software Engineer, AmpUp

In the process of evaluating tools to solve his problem, Nguyen ruled out a smaller flag vendor on staying power. "Cloudbees seems to be more established compared to the others."

Separating Deployment From Release

With Cloudbees Feature Management, AmpUp:

  • Deploys code every sprint, with each major feature behind a server-side flag
  • Keeps flag settings the same across development, QA, and production
  • Rolls features out in stages, from internal testers to specific drivers and site hosts
  • Kept GitHub Actions and the rest of its delivery tools exactly as they were
A Kill Switch for Every Time Zone

If something breaks in production, whoever is online turns the feature off in seconds, even if they didn't write the code. The engineer who owns it picks up the bug during their normal workday and turns the feature back on once it's fixed.

Flags That Shape the Product

Flags do more than gate new code. Auto refund limits sit behind a flag, so each site decides whether to skip billing for faulty sessions that deliver almost no energy. The same approach lets AmpUp tailor cost, charging behavior, and features site by site.

Measuring the Impact

Every major feature at AmpUp now ships behind a flag, and rollback is the team's first response to any incident.

  • Faster recovery: Bad features are switched off in seconds instead of waiting five to ten minutes or more for a cluster redeploy.
  • Smaller impact when something breaks: Site host features often go first to pro-plan customers and the largest sites, and driver features often start near 5% of users. The team knows the most people an issue could reach before each phase begins.
  • No overnight fire drills: Bugs found during US hours are fixed on the owning engineer's normal schedule.
  • Governance that scales: AmpUp is adding approvals to designated flags, and it logs every change.
“I like the customer obsession CloudBees has, even for the smallest of customers. Truly love the experience here."

Ganeshram Nagarajan, Head of Engineering, AmpUp

Cloudbees as Part of AmpUp's Next Phase

The keys to feature releases are now moving to AmpUp’s product managers. They define a target group once, such as the most loyal drivers, and send each new feature out one group at a time, watching how it lands before going further. The next step is automating updates and bringing in AI coding assistants to clear out old flags, so every release ships cleaner than the last

See how your team can separate deployment from release. Start free with Cloudbees Feature Management

Let’s build your path from chaos to control.

Etiam vitae pulvinar in facilisi commodo est eget. Habitasse elementum venenatis habitant tellus. Fames habitasse sed amet vitae eleifend malesuada condimentum