Manas Chittoda

• Developed a secure, scalable infrastructure for AI-based reporting, with transaction count monitoring, SNS alerts, and Lambda authorization.
• Built and deployed custom Docker images using SageMaker Notebook Instances and ECR, leveraging secrets from AWS Secrets Manager.
• Designed and deployed a public-facing, containerized budgetary estimation web app using Cloudflare tunnels, ALB, and secure S3 file handling.
• Integrated external systems, including Salesforce, Google Maps, and Azure AD, for login, mapping, and business data validation.
• Key Projects:

1.RIE – Public Budgetary Estimation Platform.

Tools: AWS ECS, ALB, Cloudflare, S3, DynamoDB, Azure AD, Salesforce, and Google Maps API.

• Designed and deployed a containerized solution (three microservices) on AWS ECS for a web-based tool used by internal teams and external channel partners.
• Integrated with Cloudflare and CloudFlareD to securely expose the tool publicly, and managed routing via the Application Load Balancer.
• Implemented secure file handling with S3 and hash-based validation, and built logic to dynamically estimate hardware needs for a given indoor layout.
• Enabled authentication using Azure AD and integrated with Salesforce to validate opportunity data, ensuring a seamless workflow for sales.

Impact:

• Empowered pre-sales and channel partners to generate accurate budgetary proposals in minutes instead of days.
• Helped scale quoting capability amid rising demand for indoor radio dot solutions.
• Closed a key gap in the sales process by enabling self-service quoting and reducing dependency on technical pre-sales teams.

2. Containerized ML Inference Platform for BCSS SOC.

Tools: AWS ECS Fargate, EFS, CloudWatch, IAM, Step Functions, Lambda, S3, API Gateway

• Built an ECS solution with three containers (Schedulers and Flask App) connected through an ALB and secured with an API Gateway and Lambda authorizer.
• Mounted EFS to persist and share inference results, stored in Pickle format, across services.

Impact:

• Reduced manual effort in bid processing by over 80%, improving operational efficiency.
• Enabled data-driven decision-making in bid selection by standardizing and automating ML workflows.

• Developed and deployed automated scripts for idle SageMaker notebook shutdown, reducing unnecessary compute usage and cutting weekly costs by 30%.
• Identified and resolved security vulnerabilities across SageMaker, VPC, and API Gateway; aligned fixes with organizational compliance standards.
• Enhanced platform reliability and safety for enterprise-wide ML use cases.