Systems we have shipped.

The challenge

An Australian sports organisation needed a customer-facing membership platform that could grow past MVP: secure accounts, media delivery at scale, and infrastructure that would survive external security review. They did not need a one-off agency handover. They needed a fractional platform team over multiple phases.

The solution

Cipher acted as the cloud and product engineering team across about eight months. We hardened authentication flows, closed four low-risk findings from an external pen test, upgraded the frontend framework to a supported release, and split a single-server backend into dedicated API and web/data tiers with a CDN in front of media streams. We also stood up observability dashboards, analytics, and an internal ops panel. Deployments and environment management ran through Ansible, with staging regression before production.

Key results

A production membership platform with tiered infrastructure, closed security findings, supported framework versions, and observability the internal team can operate. The engagement proved fractional platform engineering works: not a throw-over-the-wall build.

Technologies

• Next.js
• Node.js API
• PostgreSQL
• CDN
• Ansible
• CloudWatch
• Auth hardening
• Observability

The challenge

An Australian firm receiving large volumes of unstructured inbound material (PDFs, attachments, and correspondence) needed analysis capability without sending sensitive content to shared SaaS tools. Manual triage did not scale. They needed agents inside a boundary they owned and could audit.

The solution

We architected a secure AWS footprint with VPC isolation, role-based access, and full audit logging. Incoming material flows through Amazon SES into a processing pipeline where Hermes-class and AgentCore agents extract, categorise, and structure content into PostgreSQL. Analysts work from governed exports and dashboards, not raw inboxes. Strands orchestrates agent behaviour with guardrails on what each workflow can access.

Key results

Internal teams review structured output instead of manually opening every attachment. The environment is repeatable for future research waves: same security boundary, same agent pipeline, new intake rules as needed.

Technologies

• AWS VPC
• Amazon SES
• AWS Bedrock
• AWS AgentCore
• Strands Agents
• Hermes-class agents
• PostgreSQL
• Audit logging