A leading bank halves its cloud migration timeline using a GenAI-powered agentic ETL Crew
How OneByZero's three-agent AI system automated ETL code generation, testing, and migration execution, enabling a major bank to independently migrate 60% of its pipelines.
Powered by Amazon Bedrock and a multi-agent ETL Crew
50%
reduction in project completion time and human effort
60%
of ETL pipelines migrated independently by the client
12M
customers served by the bank
Overview
About the customer
The client is a leading bank in Southeast Asia, serving more than 12 million customers. As part of a strategic modernisation initiative, the bank undertook a large-scale migration from on-premises infrastructure to a cloud-based data architecture, a project that required rewriting hundreds of ETL scripts and analytical models to function in a fundamentally different environment. OneByZero partnered with the bank to design and deploy a GenAI-powered agentic solution capable of accelerating this migration, while simultaneously building the bank's internal capacity to continue the work independently.
The challenge
Architecture-wide rewriting at scale
The bank's move to the cloud was not a lift-and-shift. The divergence between on-premises and cloud architectures meant existing ETL pipelines and data models could not be ported directly, they had to be fundamentally rewritten and re-optimised for cloud-native environments.
01
Not a lift-and-shift
On-prem and cloud architectures diverged, so ETL pipelines and data models had to be rewritten from the ground up, not ported.
02
Hundreds of scripts to rewrite
Hundreds of ETL scripts and analytical models needed re-optimisation for a fundamentally different environment.
03
Capacity and timeline pressure
Doing the work manually would have required significant data engineering effort over an extended timeline, creating capacity constraints and delaying outcomes.
04
Quality at scale
The bank needed to encode migration logic and reduce human dependency on repetitive rewriting, without sacrificing quality or correctness across a large codebase.
The solution
The ETL Crew: three specialised AI agents working as one
OneByZero designed and deployed the ETL Crew, a GenAI-powered agentic system built around three distinct AI agents, each with a defined role, working collaboratively to handle the full migration workflow end to end.
#Agent 01
AI Tech Lead
Receives and parses instructions via email and orchestrates the full migration workflow.
- Parses natural-language migration instructions
- Coordinates handoffs between agents
- Integrates with Outlook for seamless communication
#Agent 02
AI Developer
Fetches existing code from the repository and rewrites it for the cloud.
- Retrieves existing code using natural-language prompts
- Rewrites and translates ETL scripts for cloud-native architectures
- Produces Glue-optimised PySpark and Python
#Agent 03
AI QA Tester
Conducts automated testing and independently executes migration with QA built in.
- Runs automated tests on all generated code
- Independently executes migration activities
- Quality assurance built in at every step
AWS services used
Built on AWS, available on Marketplace
The ETL Crew runs on AWS-native services for orchestration, runtime, and validation, and is available as a verified solution on AWS Marketplace.
Amazon Bedrock
Powers the multi-agent orchestration and LLM reasoning for all three AI agents via Bedrock's managed foundation model APIs.
AWS Glue
Target runtime for generated cloud-native ETL pipelines; the AI Developer agent produces Glue-optimised PySpark and Python scripts.
AWS Lambda
Handles event-driven triggers for agent handoffs, pipeline execution steps, and automated QA test runs.
Amazon SageMaker
Provides the ML model hosting and validation environment used during automated QA testing of generated ETL code.
AWS GenAIIC & Partner Agent Factory
Co-innovation programme that provided architecture patterns, GenAIIC methodology, and Strands SDK assets to shape the ETL Crew's enterprise-grade design.
AWS Marketplace
The ETL Crew is available as a verified, enterprise-ready solution on AWS Marketplace, simplifying procurement and deployment.
Co-innovated with AWS GenAIIC and delivered by OneByZero, an AWS partner, using the Strands SDK and AWS-native agentic patterns.
Outcome
Half the time, twice the autonomy
50%
less project time and human effort
60%
of pipelines migrated independently
3
AI agents working as one crew
The ETL Crew delivered measurable impact across both speed and capability. Project completion time and human effort were reduced by 50%, enabling the bank's teams to reallocate engineering capacity from repetitive migration tasks to higher-value architectural and governance work.
- Project completion time and human effort reduced by 50%
- 60% of ETL pipelines and models migrated independently by the client
- Engineering capacity reallocated to higher-value architecture and governance work
- GenAI agents embedded directly into email, Jira, and GitLab workflows
- Adoption achieved without retraining or new tooling
- A durable capability the bank continues to operate and scale on its own
About OneByZero
OneByZero is a frontier Systems Integrator specialising in AI Coworker design, build, and deployment for regulated enterprises across financial services, telecommunications, and retail. We operate across ASEAN, India, ANZ, and Japan, with a presence in the United States, combining domain expertise in regulated industries with deep AI engineering capability.
HQ Singapore · Tech HQ USA · Australia · India · Indonesia · Malaysia · Philippines · Thailand · Vietnam
