top of page
Minimalist Vases Display
Cloud-based multi-country call center architecture integrating SIP telephony with CRM for resilient customer support operations.

Cloud Migration & Infrastructure Transformation - Kasha Rwanda & Kenya

Led the multi-country stabilization and cloud migration of Kasha’s call center infrastructure, transitioning from fragile on-prem E1-based telephony to a resilient, CRM-integrated cloud model that delivered near-zero downtime and transformed customer support into a scalable growth enabler.

Context & Business Challenge

When I joined Kasha Global, the call center represented a single point of operational failure.

In Rwanda, the telephony infrastructure relied on on-prem E1 card technology hosted in an unreliable physical location, dependent on unstable power and network connectivity. Frequent outages resulted in missed customer calls, lost revenue, and deteriorating customer experience.

Kenya operated through a more stable Safaricom colocation setup, but Rwanda’s instability dragged overall performance down. Maintenance costs continued rising while reliability declined.

At the same time, the rollout of Freshworks CRM exposed the need for modern, cloud-ready telephony capable of integrating seamlessly with customer workflows.

Customer support was mission-critical particularly in a health-focused business. Continued downtime posed brand, regulatory, and growth risks.


Objective & Success Criteria

The initiative aimed to eliminate infrastructure fragility and modernize telephony into a scalable, cloud-first model.


Key objectives

  • Remove call center services from unreliable on-prem environments

  • Colocate infrastructure within professional data centers

  • Eliminate power and network-related outages

  • Migrate telephony to cloud-based SIP architecture

  • Integrate call center services with Freshworks CRM

  • Reduce ongoing infrastructure and maintenance costs


Success was defined by

  • Stable operations across Rwanda and Kenya

  • Near-zero unplanned downtime

  • Seamless CRM-linked call logging and history

  • Reduced maintenance spend

  • Architecture scalable for new markets


My Role & Ownership

I led the program end-to-end as Program Owner and Technical Lead.

My responsibilities included:

  • Infrastructure assessment and future-state design

  • Vendor engagement with AOS (Rwanda), MTN (Rwanda), Safaricom (Kenya), and later MTN (South Africa).

  • Designing phased migration from on-prem to colocation to cloud

  • SIP trunk integration strategy

  • Risk management and multi-country coordination

  • Executive reporting and stakeholder alignment


This was a high-risk, multi-vendor, multi-country modernization effort.


Program Strategy & Execution
Phase 1 - Infrastructure Stabilization (Colocation)

I began by auditing the existing telephony infrastructure in Rwanda.

Actions included:

  • Selecting AOS data center for colocation

  • Migrating call center servers from office premises to controlled data center environments

  • Coordinating redundancy for power and network connectivity

  • Eliminating location-based outages


This phase stabilized operations while preparing for cloud transition.


Phase 2 - Cloud Telephony Migration

With infrastructure stabilized abit, I initiated a cloud-first telephony strategy aligned with the Freshworks CRM rollout.

Key steps included:

  • Integrating SIP trunks with FreshCaller (Freshworks Cloud Contact Center)

  • Coordinating MTN Rwanda cloud SIP rollout

  • Working with Safaricom Kenya for seamless integration

  • Gradually decommissioning legacy hardware

  • Conducting failover testing and validation

  • Training support teams on new workflows


The migration reduced hardware dependency and introduced scalable, IP-based routing.


Technical Architecture

Architecture Evolution
  • Before: On-prem PBX with E1 cards and frequent outages

  • After: Cloud-based SIP telephony integrated directly with CRM


Technologies Implemented
  • SIP trunking (MTN Rwanda, Safaricom Kenya)

  • Freshcaller (cloud contact center)

  • Data center colocation (AOS Rwanda, Safaricom Kenya)

  • IP-based call routing and recording


Key Capabilities Enabled
  • CRM-linked call logging and history

  • Call recording and performance tracking

  • Scalable agent provisioning

  • Reduced physical infrastructure dependency


The final design prioritized resilience, scalability, and CRM-native integration.


Results & Impact

Before
  • Frequent and unpredictable downtime

  • High maintenance costs

  • Operational firefighting

  • Poor customer experience

  • No scalable model for expansion


After
  • Near-zero unplanned downtime over 12 months

  • Stable and predictable call center operations

  • Reduced infrastructure and maintenance costs

  • Seamless CRM integration

  • Scalable, cloud-first telephony architecture

  • Improved customer satisfaction and agent productivity


Strategic Impact: Customer support transitioned from a reliability risk to a stable operational backbone capable of supporting regional growth.


Key Challenges & Approach

  1. Telco Delays (Rwanda): Slow provisioning timelines created migration delays.

    • Approach: Secured executive sponsorship, instituted structured milestone tracking, and maintained persistent escalation.


  2. Legacy Hardware Instability: Old servers continued failing during transition.

    • Approach: Accelerated cloud migration to reduce dependency on physical infrastructure.


  3. Multi-Vendor Coordination Across Countries: Different vendors with varying standards.

    • Approach: Centralized architecture blueprint, clear vendor ownership boundaries, and consistent governance oversight.


What I Would Do Differently
  1. Move directly to cloud where feasible, bypassing interim colocation

  2. Negotiate stronger SLAs upfront with telcos

  3. Budget for cloud telephony earlier in planning cycles

  4. Pilot cloud migration earlier to compress timelines

  5. Review and tighten exit clauses for underperforming vendors

© 2026 by George Hillary Kafuko

Let's Connect

  • LinkedIn
bottom of page