From Broken Go Microservices to a High‑Performance DevOps Culture: A Tech Director’s Playbook

1. Pragmatic Retreat: Strategic Decisions

Facing a fragile, over‑engineered Go micro‑service architecture that caused frequent cascade failures, the author chose to abandon the existing stack and make three decisive moves:

Location strategy : Relocated the R&D center from high‑cost, high‑turnover cities to Xiamen, where talent is abundant, stable, and cost‑effective.

Technology shift : Switched the whole stack from Go to the Java ecosystem to gain richer tooling, better third‑party component support, and a larger talent pool, prioritising rapid delivery over cutting‑edge novelty.

Product‑first mindset : Re‑defined the role from CTO to chief product manager, turning the system from a mere data recorder into a business‑driving workflow engine.

2. Four‑Person Sprint: 60‑Day Rebuild

Within two months, a four‑person core team (three developers and one product manager) adopted an MVP‑first approach, stripping away non‑essential features and focusing on the most painful business processes. They lived on‑site, collaborated closely with store managers, and redesigned the database to create a coherent business loop. The resulting 1.0 version delivered sub‑second page loads and automated previously paper‑based approvals, dramatically improving user satisfaction.

3. From Heroic Overtime to Engineering Efficiency

As the team grew, the author introduced systematic engineering practices to replace ad‑hoc heroics:

Standardised demand management : Established a backlog with business value assessment, technical review, and priority ranking to ensure the team works on the highest‑impact items.

Automated CI/CD pipeline : Built a Jenkins‑driven pipeline integrated with Gitee (SaaS version) so that code commits automatically trigger builds, unit tests, and deployments, eliminating manual packaging errors.

Metrics‑driven improvement : Tracked lead time, defect density, and deployment success rate, using the data to pinpoint bottlenecks such as long testing cycles or unclear requirements.

These practices cut deployment friction, accelerated iteration speed, and aligned business and engineering through transparent metrics.

4. Crossing Physical Boundaries: Digital Teaching Breakthrough

To digitise a hands‑on education business, the team adopted an electromagnetic handwriting board that preserved the feel of writing on paper while capturing high‑frequency stroke data. The backend reconstructed the trajectory on a Canvas, automatically scoring maze‑solving tasks, thus turning a manual, low‑throughput process (1 teacher per 6 students) into an automated, scalable system.

Subsequent work added real‑time audio‑video streaming, enabling an OMO (online‑offline) model that reduced centre footprint and multiplied teacher productivity. The amassed data opened possibilities for BI‑driven personalised teaching.

By 2018, the team began experimenting with AI, training CNN models on handwriting trajectories and video streams to recognise micro‑expressions and assess focus, laying the groundwork for intelligent, data‑driven education.

Conclusion: The Dual Role of a Tech Director

The author reflects that a tech leader’s highest value lies not in chasing the latest buzzwords but in closing the loop between technology and business outcomes. Effective leadership means being a strategic adviser to senior management while acting as a mentor and enabler for the engineering team, fostering a culture where responsibility, trust, and continuous improvement drive sustainable growth from 0 to 1 and beyond.