DAKSH is engineered for deployment in secure, cost-optimized, and scalable environments, leveraging a modern serverless architecture to eliminate the overhead associated with traditional infrastructure management. Its backend is composed of decoupled services, each performing a specialized role in the overall pipeline — from query intake and preprocessing to context retrieval, response generation, and audit logging.
At the entry point, API Gateway serves as the request router, managing incoming queries from web, mobile, or integrated platforms. It handles authentication, rate limiting, and request validation, acting as the secured gateway into the DAKSH ecosystem. Downstream, AWS Lambda-based microservices execute discrete tasks such as input validation, vector embedding lookup, retrieval orchestration, and model inference. These stateless functions automatically scale with user load and terminate after execution, significantly reducing idle compute costs while maintaining high concurrency support.
For data persistence and structured storage, Amazon S3 is used to manage user-uploaded documents and processed knowledge chunks, while RDS (Relational Database Service) or DynamoDB stores metadata including document mappings, session logs, and schema references. All stored data is encrypted using AWS KMS (Key Management Service), ensuring confidentiality and regulatory alignment with GDPR, ISO 27001, and DPDP standards.
DAKSH integrates CloudWatch for logs, event tracing, and performance insights, while optional integration with Prometheus and Grafana provides real-time observability for latency, throughput, and vector search performance. To enforce strict access controls, DAKSH employs IAM roles and policy-bound service permissions, ensuring each microservice can access only the resources it requires.
This architecture not only supports auto-scaling during peak usage but also enables clear responsibility segregation by function — isolating sensitive services like inference and audit logging from public endpoints. Combined with token-based request authentication and fully encrypted data paths, DAKSH delivers a deployment model that is robust, efficient, and compliant by design.