Scaling Bykea to 10M+ Users & 1.5M Monthly Trips

Co-Head of Engineering / Deputy Head of Engineering · 2023 · 4 min read

Scaled the platform from a fragile QA-dependent model to a quality-embedded engineering culture — delivering 500+ releases, 99.8% infrastructure uptime, and $240K boost in recurring revenue.

Overview

Bykea is Pakistan's largest ride-hailing and logistics super-app. After leading a multi-year evolution from QA Engineer to Deputy Head of Engineering, I drove the platform's reliability, quality, and team transformation as it scaled from 3M to 10M+ users and grew to orchestrate 1.5M trips monthly.

Problem

As Bykea grew aggressively, the traditional QA-at-the-end model was breaking down. Releases were slow, production bugs were high, and the QA team was a bottleneck rather than an enabler. The platform needed to scale both technically and organizationally.

Constraints

  • High-stakes consumer app: crashes affected millions of rides and courier deliveries
  • Fast-moving startup with aggressive release cadence
  • Distributed teams with varying levels of QA maturity
  • Limited automation coverage creating regression risk at scale

Approach

Transformed QA from a gate-keeping function into embedded quality assistance. Introduced automation-first strategy, shifted testing left, and rebuilt the squad model around cross-functional ownership. Instrumented the platform with deep observability and established release standards that teams could independently meet.

Key Decisions

Shift from Quality Assurance to Quality Assistance model

Reasoning:

Traditional QA was creating a handoff culture where devs threw code over the wall. Moving to quality assistance embedded QA thinking into every sprint, making engineers co-owners of quality and reducing bug leakage at the source.

Alternatives considered:
  • Hire more QA engineers for faster review
  • Introduce automated gates in CI only
  • Outsource regression testing

Appium + Python for mobile automation, Rest Assured + Postman for API layer

Reasoning:

Appium gave us true cross-platform mobile coverage; Python scripts were maintainable by the team. The API layer needed to be validated independently of the UI to catch integration regressions faster.

Alternatives considered:
  • Espresso / XCUITest (platform-specific)
  • Manual regression only
  • Third-party testing services

Mandatory ISTQB certification for the QA team

Reasoning:

Standardizing on a globally recognized certification ensured the entire team shared a common testing vocabulary, principles, and practices — reducing inconsistency in test coverage and bug reporting quality.

Alternatives considered:
  • Internal training only
  • Role-specific certifications

Tech Stack

  • Appium
  • Python
  • Java
  • JMeter
  • Postman / Rest Assured
  • GitLab CI/CD
  • JIRA
  • Confluence
  • BrowserStack
  • Cucumber / Gherkin

Result & Impact

500+ Successful production releases managed
10M+ Active users supported on the platform
1.5M Monthly trips orchestrated
$240K Boost in annual recurring revenue
99.8% Infrastructure uptime maintained
99.6% Crash-free rate across all mobile apps
50% Reduction in QA cycle time
20% Decrease in production bug leakage

The Challenge

When I joined Bykea in 2019 as a Software QA Engineer, the platform was serving ~3M users. By 2022, user numbers had tripled and the engineering org had grown significantly — but the quality model hadn’t kept pace.

Releases were stressful. Regression runs were long. And the QA team was constantly the last line of defense, catching issues that should have been caught much earlier.

What I Built

Automation-First Foundation

I architected a mobile automation framework using Appium + Python from scratch, covering the customer Android app, customer iOS app, and partner app. The framework ran on Jenkins with BrowserStack for real-device coverage.

For the API layer, I wrote comprehensive test suites in Postman and JMeter that were integrated into the CI pipeline — so every backend deployment was automatically regression-tested before any UI testing began.

Result: 90%+ test coverage. Regression testing time cut by ~50%.

Quality Assistance Transformation

The cultural shift was harder than the technical work. I restructured QA processes so that:

  • QA engineers were embedded in squads, not in a separate team
  • Test planning started at the Epic/User Story review phase, not post-development
  • Developers owned unit and integration tests; QA focused on system, E2E, and exploratory
  • Post-mortems became standard practice to learn from production issues

Result: QA time reduced by 50%. Production bug leakage dropped by 20%.

Release Readiness Framework

I defined and enforced release readiness criteria across all three apps (customer Android, customer iOS, partner Android). This included sign-off checklists, automated smoke test gates, and rollout monitoring dashboards.

Result: 99.6% crash-free rate maintained at scale.

Engineering Leadership

By late 2022, I was leading 3 agile squads of 25+ engineers as Principal SQA Engineer, then Deputy Head of Engineering. My leadership work included:

  • Daily standups, sprint planning, retrospectives, and post-mortems
  • Stakeholder ETA management and epic communication
  • Talent acquisition (QA and engineering interviews)
  • Organizing community events like Product Soch Cafes
  • Driving product innovation: bidding feature, tipping, missed-call authentication

Lessons Learned

  1. Quality is a team sport. The single highest-leverage change was making developers feel ownership over quality — not just shipping features.

  2. Automation ROI compounds. Early investment in a solid framework paid dividends across every subsequent release cycle.

  3. Leadership is process design. At scale, how you run ceremonies, handle decisions, and communicate priorities matters more than individual technical skill.

All projects