Aivar designed and implemented a serverless, event-driven architecture focused on Intelligent Automation, Infrastructure as Code, and Generative AI workflows.

• Infrastructure as Code (IaC) & Automation: The entire AWS environment is provisioned using Terraform. Aivar automated the deployment of VPCs, AWS Lambda functions, SQS queues, and DynamoDB tables, ensuring the infrastructure is versioned and reproducible across environments.
• Modern GitOps & CI/CD Pipeline: Aivar implemented a GitOps-driven CI/CD workflow using GitHub Actions and AWS CodeDeploy. The pipeline automates packaging, executes security scanning (Checkov, Trivy, Bandit), and manages controlled deployments with traffic shifting and automated rollbacks.
• AI-Powered Architecture & Intelligent Extraction: The solution leverages Amazon Bedrock (Claude 3.7 Sonnet) and Amazon Textract for advanced multi-stage data extraction. This allows the system to analyze PDFs page-by-page, identify invoice subsets, and convert unstructured text into structured JSON for downstream integration.
• Event-Driven & Scalable Workflow: The system handles high volumes using an event-driven flow triggered by Amazon SES and S3. Asynchronous processing is managed via SQS queues and Lambda functions to decouple services, while AWS X-Ray and CloudWatch provide end-to-end observability.
• Security & Governance: Security is embedded via a "least-privilege" IAM model, KMS encryption for data at rest, and AWS Signer for code integrity.

• Validated Scalability: Successfully validated the platform's ability to handle 1,000+ invoices per hour.
• Operational Excellence: Achieved fully automated, deterministic deployments, reducing the build time from 60 minutes to 6 minutes—a 10x improvement.
• High Extraction Accuracy: Established an AI-powered extraction workflow with a targeted accuracy rate of >95%.
• Cost Efficiency: Minimized Total Cost of Ownership by utilizing a serverless model, resulting in a 60% reduction in non-production environment costs through the use of ephemeral environments.
• Accelerated Release Cycles: Improved release frequency from quarterly to weekly deployments by standardizing prod and non-prod environments with IaC.

Aivar successfully delivered a scalable, secure, and operationally resilient foundation for the client’s AI-driven freight operations. By leveraging serverless architecture and advanced Generative AI models, the solution transformed a manual, fragmented workflow into a highly efficient, automated system. This implementation ensures that the resulting Minimum Viable Product provides a future-ready platform for sustained innovation and measurable business value.

‍