From GCP to AWS: How Mining Terminal Modernized Infrastructure and Scaled AI Workloads
The Client
Mining Terminal is a mining intelligence and research platform focused on aggregating and structuring technical mining data from public filings, feasibility studies, drill reports, and regulatory disclosures. Covering over 2,300 mining companies and 17,000 mining projects across more than 420,000 technical documents, the platform delivers structured datasets, APIs, and project screening tools for institutional investors, analysts, and industry professionals.
Part of the AI-driven data aggregation ecosystem, Mining Terminal processes massive volumes of unstructured geological, drilling, and regulatory documents through automated extraction pipelines.
- Industry:Research Services / Mining Intelligence & AI
- Company Size:5 employees
- Country:Germany
Challenges
Mining Terminal was previously running its AI document processing infrastructure on Google Cloud Platform (GCP), which presented several limitations:
- Scalability constraints – GCP’s tiered model access system restricted inference capacity, creating severe bottlenecks for large-scale document ingestion and analysis.
- High operational costs – Manually managed infrastructure demanded constant hands-on effort from a single founder, with no clear cost control mechanisms in place.
- Limited flexibility – Heavy reliance on manual provisioning and a lack of infrastructure-as-code made expanding coverage across new mining projects difficult.
- Need for a robust setup – Required high-throughput, low-latency foundation model inference, automated high availability, and enterprise-grade security to prevent document processing delays.
Solutions
To solve these challenges, Cloudvisor executed a phased migration of Mining Terminal’s core application workloads and AI extraction pipelines from GCP to AWS.
Phase 1: AWS Foundation & Security Baseline
- Pre-configured a 3-tier, 3-AZ VPC architecture (public, private-app, private-data) following security best practices.
- Deployed an enterprise-grade security baseline—CloudTrail, GuardDuty, Secrets Manager, and IAM least-privilege with MFA.
Phase 2: Scalable Compute & AI Model Integration
- Replatformed application servers to EC2 Auto Scaling Groups for self-healing, burst extraction capacity.
- Integrated Amazon Bedrock (Nova Pro, Claude) via private VPC Endpoints for rate-limit-free AI extraction.
Phase 3: Database & Analytics Modernization
- Migrated the primary database to Aurora PostgreSQL Serverless v2, scaling automatically down to a minimal compute floor.
- Configured Athena and Glue Data Catalog for serverless SQL analytics on S3-stored extraction outputs.
Phase 4: CI/CD Automation & Knowledge Transfer
- Automated deployments via GitHub Actions with passwordless AWS OIDC authentication.
- Codified 100% of infrastructure in Terraform and delivered hands-on knowledge transfer workshops.
AWS Services Used
- Amazon EC2 Auto Scaling Groups
- Amazon Aurora PostgreSQL Serverless v2
- Amazon Bedrock
- Amazon Athena
- AWS CloudTrail
- Amazon S3
- Amazon SQS
- AWS Secrets Manager
- Amazon GuardDuty
- AWS Glue Data Catalog
Results
The migration empowered Mining Terminal with a modern AWS setup that eliminated inference limits, improved reliability, and unblocked business growth — all achieved without disrupting ongoing operations.
100% funded by AWS
Fully funded through the AWS MAP program — zero upfront migration cost for the customer.
- Optimized cloud spend
Aurora Serverless v2's minimal scaling floor and auto-scaling extraction instances cut idle compute costs.
-
Unblocked business growth
Eliminated GCP rate limits, enabling on-demand extraction across 420,000+ filings and unblocking 12 waiting customers.
-
Standardized, faster deployments
Terraform and GitHub Actions CI/CD cut new environment provisioning to under one business day.
