Case Studies | SaaS B2B

From Railway to a purpose-built AWS platform

Railway to AWS Migration

About

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

Challenge

The migration had four focus areas, each tied to a constraint of the existing Railway environment.

Persistent asset storage

Generated HTML outputs, product images, and brand videos were written to Railway's ephemeral filesystem, which put per-user assets at risk on every redeploy. The client needed durable object storage with a global delivery path so that finished work stayed available and loaded quickly for users anywhere.

A scalable, durable data layer

PostgreSQL and Redis on Railway offered limited headroom and backup control. The platform needed managed, highly available data services with automated backups and point-in-time recovery, while preserving all rate-limiting, session, and circuit-breaker behaviour with no regression.

Inference cost and control at scale

Inference ran through direct Anthropic Claude API calls. As usage grew, the client needed AWS-native inference with the existing fallback routing intact, plus a credible path to reduce per-token cost as the user base expanded.

Runtime resilience and a downtime-free cutover

The application ran as a shared runtime with git-push deploys. The client needed an orchestrated, auto-scaling runtime, a proper CI/CD pipeline, and validated real-time streaming, all delivered through a phased cutover that avoided disruption to active users.

Solution

Cloud Combinator sequenced the work so that each layer could be migrated, validated, and cut over independently, keeping the live platform stable throughout.

15

Specialist AI agents running in parallel on AWS

4

Phase migration delivered without user downtime

60-80%

Projected inference cost reduction at 5,000+ users (projected)

By the numbers:

  • 15 - Specialist AI agents running in parallel on AWS
  • 4 - Phase migration delivered without user downtime
  • 60-80% - Projected inference cost reduction at 5,000+ users (projected)
Changes

The client now operates end-to-end on AWS with no dependency on Railway. Every one of the fifteen specialist agents executes correctly within the AWS environment, real-time streaming holds up under Fargate networking, and the data layer runs with automated backups and recovery in place. Acceptance was measured against the migration of every layer with data integrity preserved and live production traffic running stably on ECS Fargate.

  • Durable asset storage — Amazon S3 with CloudFront replaced Railway's ephemeral filesystem, so generated sites, images, and video persist and load quickly worldwide.
  • Managed data layer — RDS PostgreSQL (Multi-AZ) and ElastiCache for Redis provide high availability, automated backups, and point-in-time recovery with no behavioural regression.
  • AWS-native inference — Amazon Bedrock (Claude) now serves inference through IAM, with OpenAI and Google Gemini fallback preserved end-to-end.
  • Resilient runtime — ECS Fargate behind an ALB with auto-scaling and a CI/CD pipeline replaced the shared runtime, and the Railway services were decommissioned.
  • Handover — CloudFormation and SAM templates, operational runbooks, and working sessions were delivered as an architecture handover pack.

With the core platform live on AWS, the Phase 4 scale layer positions the client to move inference onto self-hosted SageMaker capacity and, in time, a fine-tuned model, with a clear path toward a Well-Architected Framework Review and Foundational Technical Review as the platform grows.

AWS Stack

Amazon S3

For durable per-user workspace storage.

Amazon CloudFront

For global, low-latency asset delivery.

Amazon Bedrock

For managed Claude inference with retained fallback routing.

Amazon RDS

For PostgreSQL for a highly available, backed-up data layer.

Amazon ElastiCache

For Redis for rate limiting, sessions, and circuit-breaker state.

AWS Fargate

On Amazon ECS for the auto-scaling application runtime.

Amazon SageMaker

For self-hosted GPU inference in the scale phase.

AWS CloudFormation

And SAM for reproducible infrastructure as code.

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.