eEdenic
← Selected work

Platform · Overhaul

An 18-service microservices overhaul.

We took a stalled platform migration and drove it to production — clean service boundaries, real-time data, full observability, and a handoff the in-house team kept building on. Four weeks, audit to cut-over.

Node.jsGoRustPythonPostgreSQLPostGISRedisClickHouseKafkaS3
18

autonomous services

12 / day

prod releases

99.985%

uptime

4 weeks

audit to cut-over

Snapshot

Scope

18 autonomous services, 320+ Kubernetes objects, 400+ Git commits

Scale

89 white-label companies, 2.4M annual trips, 48K drivers online at peak

Velocity

12 prod releases per day with a <10 min rollback window

Reliability

99.985% uptime, zero critical incidents since go-live

Starting point & pain

  • A tangled monolith blocking independent releases across 89 tenants
  • Deploys risky and slow — every change touched everything
  • No unified observability across a fleet-scale, real-time system
  • Scaling costs climbing faster than the business

Objectives

  • Decompose into independently deployable, tenant-aware services
  • Ship 10+ times a day safely with instant rollback
  • Instrument the whole system — metrics, logs, traces, SLOs
  • Harden security and governance across the pipeline

Transformation roadmap

  1. Phase 1

    Audit & service boundaries

    Map the domain, carve service boundaries, agree the target architecture.

  2. Phase 2

    Platform & IaC

    Kubernetes platform, GitOps pipelines, and infrastructure as code.

  3. Phase 3

    Service migration

    Extract and migrate services in parallel with zero-downtime cut-over.

  4. Phase 4

    Data & ML pipeline

    Streaming data with Kafka + ClickHouse, ML-ready feature flows.

  5. Phase 5

    Observability & handoff

    SLOs, dashboards, runbooks, and a documentation bundle the team owns.

Engagement overview

Duration

4 weeks from audit to cut-over

Team

3 DevOps engineers, 2 backend specialists, 1 SRE, 1 project lead

Deliverables

Architecture, IaC, CI/CD, service code, observability stack, 350-page docs

Key takeaways

  • Independent deploys turned a quarterly release into a daily habit
  • Observability-first meant incidents were caught before customers noticed
  • A clean handoff let the in-house team keep shipping without us

Have a migration that stalled?

We embed, drive it to production, and hand it back clean. Talk to a founder.

Talk to a founder ↗