Multi-Tenant SaaS Architecture & Platform Engineering
I design and build multi-tenant platforms where tenant isolation, role-based permissions, approval workflows, and audit trails are enforced in the architecture - not left as conventions your team has to remember at every call site.
Core Capabilities
1. Tenant Isolation Enforced At The Data Layer
Tenant scope applied in the data-access layer rather than remembered by each developer at each call site, so a new query inherits it instead of needing to re-add it. I have shipped both JWT-claim-based isolation and per-tenant database routing, and the right choice depends on your compliance posture, not on fashion.
2. Role-Based Access Control With Real Hierarchies
Most RBAC implementations fail the same way: a permission grants an action but carries no notion of whose data the caller may act on, so any tenant's admin can reach across the boundary. I model permissions against the organisation's actual hierarchy and enforce them server-side, where the UI cannot be the only thing standing in the way.
3. Approval & Maker-Checker Workflows
Financial and regulated systems need a second pair of eyes on a change, not a trust-the-operator model. I build staged approval pipelines - draft, review, sign-off, with a return path for rejections - with the transition rules enforced server-side, so a stage cannot be skipped by calling the API directly.
4. Action-Level Audit Logging
A durable record of who did what, to which record, when, and from where. Built in as infrastructure rather than bolted on per feature, so new functionality is auditable by default and your team is not reconstructing history from application logs during an incident.
5. Per-Tenant Theming & Configuration
One codebase serving multiple client organisations, each with its own branding, products, and configuration - without forking per tenant. Theme-aware component libraries with tenant-scoped configs keep onboarding a new client a configuration change rather than a release.
6. Performance At Realistic Data Volumes
Platform tables grow faster than anyone plans for. Virtualised data grids, server-side pagination and filtering, and query patterns that stay flat as records accumulate - so the admin screens that worked in demo still work two years in.
Why This Is Worth Getting Right Early
Multi-tenancy is an architectural decision, not a feature. The cost of getting it wrong is rarely a visible bug - it is a quiet one, discovered by a client rather than by your test suite, in a system where every endpoint already assumes the wrong model.
- Tenant-boundary bugs are silent: the system looks correct until one client sees another's data
- Permission models bolted on after launch require touching every endpoint you already shipped
- Approval workflows enforced only in the UI are not enforced at all
- Audit requirements usually arrive from compliance after the architecture is already fixed
- Retrofitting isolation into a single-tenant codebase costs more than building it in from the start
Technologies & Stack
In use on platforms I have led or built end to end:
See It In Practice
Bancassurance Platform
Multi-tenant insurance SaaS for a financial services provider. I led the frontend architecture: JWT-based tenant isolation, action-level audit logging, and a maker-checker approval workflow across 7 entity types, enforced from the UI through to the API.
Steward
Financial governance platform modelling a real approval hierarchy - draft, finance review, chair sign-off, with a needs-changes return path - over Prisma and PostgreSQL with role-scoped permissions throughout.
WarehouseLens
Multi-tenant warehouse platform built solo: a two-level identity model where a platform tenant provisions the others, 6 roles across 21 permissions, and scope enforcement that reaches individual warehouses - with cross-tenant and self-lockout guards under test.
How We Can Work Together
Contract
Fixed-scope platform work - building the multi-tenant foundation, or a permission and audit layer over an existing system.
Consulting
Architecture review before you commit: tenancy model, permission design, and where your current approach will bite.
Full-time
Embedded on your engineering team owning the platform layer long-term.
Based in Nairobi, Kenya - available for remote and on-site engagements across East Africa and globally.
Building a platform that has to serve many clients?
Whether you are designing the tenancy model from scratch, retrofitting permissions into a system that shipped without them, or want a second opinion before you commit - let's talk through it.
Get in touch