Case Studies | FinTech

Re-platforming the platform from Google Cloud to AWS

GCP to AWS Migration

About

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

Challenge

The challenge had four focus areas that shaped the whole engagement.

Breaking single-cloud dependency

The platform's entire stack sat on GCP, from storage and identity to the AI models themselves. Moving to AWS meant finding a native equivalent for every component and re-mapping ten distinct GCP services without leaving functional gaps, so the product would run the same or better once the move was complete.

Preserving AI quality

The AI layer was the most critical part of the migration. Vertex AI and Gemini handled language, embeddings, vision and OCR, vector search and grounded web answers. Every one of these had to be rebuilt on Amazon Bedrock and Claude and validated against the existing Gemini benchmarks, because a financial assistant that answers less accurately after a migration is a step backwards for its members.

Seamless authentication

The platform uses Google sign-in only, so the identity move from Firebase to Amazon Cognito had to be invisible to users. That meant a bulk migration keyed on each user's Google sub, no password migration, and the existing Google credentials reused, so members would keep signing in exactly as before with no reset and no friction.

Data integrity at cutover

Every document, conversation and record had to arrive intact. Cloud Firestore collections had to be re-modelled for Amazon DynamoDB and application data re-keyed from Firebase UID to Google sub, all verified by row counts, spot checks and hash comparisons so that cutover carried a target of zero data loss.

Solution

We delivered the migration in seven sequential phases. Each phase produced a working, verifiable deliverable, GCP and AWS ran in parallel through the transition, and a rollback path to GCP was maintained until the platform was fully validated on AWS.

7

Sequential phases, with parallel running and rollback at each step

Zero

Data-loss target at cutover, verified via checksums and hashes

10

GCP services re-platformed to native AWS equivalents

By the numbers:

  • 7 - Sequential phases, with parallel running and rollback at each step
  • Zero - Data-loss target at cutover, verified via checksums and hashes
  • 10 - GCP services re-platformed to native AWS equivalents
Changes

The engagement gives the client a complete, like-for-like path from GCP to AWS across all ten services in the stack, delivered in seven reversible phases and accepted once every GCP capability is demonstrated working on its AWS equivalent. Parallel running and a maintained rollback path meant the move could proceed without betting the product on a single switch.

  • Storage — All documents moved from Google Cloud Storage to encrypted, versioned Amazon S3, with integrity confirmed by checksum.
  • Authentication — Firebase Google sign-in re-created on Amazon Cognito with an ALB gateway, bulk user migration and no password reset for members.
  • AI layer — Vertex AI and Gemini workloads rebuilt on Amazon Bedrock and Claude, with embeddings, OCR, vector search and grounded web answers validated against the previous baseline.
  • Deployment — Cloud Run containers replaced by AWS Fargate on ECS with auto-scaling, behind an ALB with Cognito authentication and AWS WAF.
  • Handover — Documentation, a reference architecture and a knowledge-transfer session leave the client's team able to operate and iterate on the platform independently.

With the platform running on an AWS-native foundation, the client is set up to scale as its membership grows, on a stack projected to support an userbase expanding into the hundreds of thousands over its first three years. Multi-model routing on Bedrock and prompt caching are built in as cost controls from day one, so the platform is no longer tied to a single cloud but optimised to support whatever the client builds next.

AWS Stack

Amazon S3

For encrypted, versioned document storage replacing Google Cloud Storage.

Amazon Cognito

With an Application Load Balancer for Google-federated sign-in and access control.

Amazon DynamoDB

For low-latency operational data with Streams and Point-in-Time Recovery.

Amazon Bedrock

With Claude for conversational reasoning, vision and streamed responses.

Amazon Titan Embeddings

And Amazon OpenSearch Serverless for semantic search and retrieval.

Amazon Textract

For structured OCR and table extraction from member documents.

AWS Fargate

On Amazon ECS for auto-scaling backend containers, with Amazon ECR for images.

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.