Mamun Khan

Designed and implemented a fully automated pipeline to process audio, video, and document files from user platforms such as Slack, Teams, and SharePoint. The solution leverages AWS Bedrock to extract insights, generate summaries, and produce transcripts from uploaded media.

The pipeline uses Amazon AppFlow or API Gateway to ingest files, which are then stored in categorized S3 buckets (audio, video, and document). A lambda function triggers the creation of processing projects based on file type, with metadata stored in DynamoDB. Processing results are analyzed using AWS Bedrock, and outcomes are stored in result buckets. The system uses Amazon EventBridge for orchestration and failure handling, sending email notifications for success or error events.

The project streamlined content analysis workflows, enabling scalable and serverless media intelligence with minimal manual intervention. It provided fast, reliable insight extraction and reporting for enterprise users.

Designed and developed a domain-specific intelligent chatbot to automate customer support and call routing using Amazon Lex, Amazon Connect, and Amazon Bedrock Knowledge Base. The chatbot is capable of answering user queries based on internal company documentation and FAQs, leveraging Bedrock's RAG-based capabilities for real-time, context-aware responses.

Integrated with Amazon Connect, the chatbot intelligently routes calls to the appropriate departments based on customer intent and need. The solution significantly reduces support workload and improves customer experience by offering fast, accurate answers and dynamic call flow automation.

The system includes intent recognition, fallback handling, escalation logic, and integration with external APIs to enhance interaction quality. All components are deployed in a scalable, serverless architecture.

Designed and deployed a scalable, containerized website hosted on Amazon ECS using AWS Fargate. The entire infrastructure was provisioned using Terraform and integrated with a GitLab CI/CD pipeline for automated build, testing, and deployment.

The architecture features a high-availability VPC with public and private subnets across multiple availability zones, an Application Load Balancer (ALB), and ECS services running on AWS Fargate. Domain routing was handled via Amazon Route 53.

The GitLab pipeline automated the end-to-end deployment process using AWS CodePipeline, CodeBuild, CodeDeploy, and Amazon ECR. Additional integrations included Amazon API Gateway and AWS Lambda for back-end interactions, and Amazon SES for email functionality.

This project demonstrates a production-grade, cost-efficient, and fully automated approach to deploying and managing web applications in the cloud.