Case Studies | SaaS B2B

Containerising an AI compliance pipeline for broadcast media

Pipeline Containerisation

About

A SaaS B2B business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The challenge had four focus areas, each rooted in the constraints of the original monolithic deployment.

A monolithic, single-host architecture

The full compliance pipeline ran on one GPU EC2 instance per tenant. This concentrated risk on a single host, mixed unrelated concerns, and made it impossible to scale ingestion, GPU processing and report delivery independently according to their very different resource profiles.

No pipeline orchestration

With the pipeline executing as one process, there was no built-in retry logic, structured error handling or execution visibility. A failure at any point in a long-running media job was difficult to isolate and recover, which is costly when video files run to gigabytes each.

Slow, manual tenant provisioning

Standing up a new tenant meant reproducing a bespoke infrastructure silo. Without parameterised infrastructure as code, onboarding was slow and error-prone, and difficult to keep consistent across tenants and regions.

Limited observability

The monolith offered little insight into container health, stage-level performance or failure conditions, making it hard to prove throughput parity or respond quickly to incidents in a compliance-critical workload.

Solution

Cloud Combinator structured the engagement as a phased delivery, opening with a discovery gate so that containerisation feasibility was proven before any build commenced, and keeping the existing EC2 deployment untouched until an intentional cutover.

3

Independent containerised pipeline stages, orchestrated by Step Functions

Under 30

Target time to provision a new tenant from a single command (projected)

9

Delivery milestones from discovery through cutover and handover

By the numbers:

  • 3 - Independent containerised pipeline stages, orchestrated by Step Functions
  • Under 30 - Target time to provision a new tenant from a single command (projected)
  • 9 - Delivery milestones from discovery through cutover and handover
Changes

The engagement defines acceptance against a clear set of success criteria: functional parity with the current EC2 deployment across all test media, throughput that meets or exceeds the EC2 baseline at production volume, strict tenant isolation, automated provisioning, live observability, and a completed operational handover. The following figures are the agreed delivery targets for the modernised platform.

  • Independent, decomposed stages — Ingestion, GPU processing and delivery run as separate ECS tasks with their own task definitions, allowing each to scale to its own resource profile rather than sharing one host.
  • Orchestrated and resilient — The Step Functions state machine adds retry logic, error handling and execution visibility across the full S3 to GPU to S3 workflow, so failures can be isolated and recovered.
  • Repeatable tenant provisioning — Parameterised CDK and CloudFormation templates target new-tenant deployment in under 30 minutes from a single command, replacing bespoke per-tenant silos.
  • Production-grade observability — Container Insights, health-check alarms and a consolidated dashboard provide the operational visibility a compliance-critical workload requires.
  • Handover — Deployment, rollback, operational and disaster-recovery runbooks are delivered alongside a hands-on workshop covering ECS operations, tenant provisioning, CI/CD and incident response.

With the pipeline containerised, orchestrated and provisioned as code, the client is positioned to onboard new tenants faster, scale each processing stage to demand, and extend the same pattern to additional regions as its compliance business grows.

AWS Stack

Amazon ECS

(Fargate and EC2) for a mixed launch-type model running the ingestion, GPU processing and delivery stages.

AWS Step Functions

For orchestrating the three-stage pipeline with retries, error handling and execution logging.

Amazon ECR

For container image storage with lifecycle policies and vulnerability scanning.

Amazon Bedrock

(Anthropic Claude Haiku 4.5) for managed language model capability within the compliance pipeline.

Amazon S3

For per-tenant media ingestion and compliance report delivery, with S3 events initiating the pipeline.

Amazon CloudWatch Container Insights

For metrics, centralised logs, alarms and the operational dashboard.

AWS KMS

For per-tenant encryption keys that maintain tenant isolation at rest.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

  • Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.
  • Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen. AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.
  • The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

We use cookies

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.