Security
Shared access needs clear technical boundaries.
ShelfGrid is a multi-tenant B2B SaaS application with tenant-scoped data access, private cloud infrastructure and explicit shared-access controls.
This page describes the current architecture supplied for security review. It does not claim certifications that ShelfGrid does not hold.
Core controls
Isolation first. Permission second. Audit where decisions matter.
Tenant-scoped data access
Every tenant-owned record is automatically scoped to the authenticated user's organisation at the data-access layer, with a deliberately explicit unscoped bypass for bounded administrative use.
Explicit supplier grants
Supplier-side access is gated by active grant records and funnels through dedicated scoping logic rather than broad cross-tenant visibility.
HTTPS and secure session controls
HTTPS is terminated at the reverse proxy with automatic certificate renewal; secure cookies, frame denial and content-type/referrer protections are enabled.
Private infrastructure
RDS is not publicly accessible, database access is limited by security groups and server administration uses AWS Systems Manager rather than an open SSH port.
Authentication and permissions are separate from cross-company access.
Web users authenticate through Django sessions; API access uses JWT. Role-based permissions control application access, while cross-organisation supplier access requires a separate explicit grant.
Password validators, brute-force protection and rate limiting protect login/token flows.
Application navigation and access follow roles and named permissions; API defaults require authentication.
Request-scoped tenant context automatically scopes tenant-owned ORM queries.
Protect sensitive credentials without overstating blanket encryption claims.
Connector credentials are encrypted with Fernet and support key rotation. Application and infrastructure secrets are injected through the environment and are not stored in source control.
Encrypted at rest in the application using Fernet.
Environment-injected; production environment storage is private and IAM-mediated.
ORM use is standard; raw analytics SQL paths are parameterised.
AWS-hosted, infrastructure-as-code and private by default.
The current deployment uses AWS eu-south-1 with EC2, RDS MySQL, S3, ECR and a private VPC. Terraform defines the environment reproducibly.
Private VPC; RDS not public; database port limited to the application security group.
AWS SSM Session Manager is used instead of open SSH.
RDS automated backups with point-in-time recovery and deletion protection.
Private ECR images, versioned deployments and Terraform-managed infrastructure.
Security is treated as an engineering discipline, not a badge.
The current architecture includes targeted tenant-isolation tests and a documented hardening backlog. Detailed technical status can be shared during an appropriate customer security review.
No certification claims are made here.
This page describes implemented architecture and controls. It intentionally does not imply SOC 2, ISO 27001 or another certification unless and until such status is formally achieved.
Need a technical security discussion?
Bring the architecture, data-flow and access questions that matter to your review.