Unity

Team Unity enables safe, policy-aware chatbot deployment using AWS—ensuring compliance, adaptability, and secure AI interaction across regions and use cases.

Comment

About the Project

We are Team Unity from Chennai, India — Toby, Prajol, Adnan, and Shantanu. With Toby and Prajol bringing in strong prior experience in Generative AI, our team set out to build a governance-first chatbot pipeline that ensures safety, adaptability, and compliance with regional data protection laws.

Inspiration

We were inspired by the need for safer AI adoption in organizations. While chatbots are becoming widespread, ensuring they meet compliance standards like GDPR, HIPAA, and India's DPDP Act without manual oversight is still a huge challenge. We wanted to solve that gap.

How We Built It

Our architecture is built fully serverless using AWS, integrating Bedrock, Lambda, DynamoDB, API Gateway, S3, and SageMaker. The system includes input/output sanitizers, region-aware policy enforcement, advisory response agents, feedback loops for model tuning, and full audit logging.

Importantly, we built and tested this pipeline entirely without relying on any free cloud credits. That forced us to design resource-conscious systems and optimize every component to be cost-efficient and production-grade.

Challenges We Faced

  • Building a modular pipeline that maintains latency guarantees
  • Ensuring dynamic policy enforcement based on user region
  • Designing a feedback loop for periodic batch retraining
  • Making everything scalable and usable without breaking compliance boundaries

What We Learned

We learned how to take governance principles and turn them into a real, working AI system with auditable processes. We also realized the value of simplicity when working without cloud credits — it made our solution lean, efficient, and practical for real-world deployment.

Built With

Share this project:

Updates