ADAAS - Artificially Driven Assistant for Automated Solutions

AI Flutter-based HR assistant with a Node/Express backend for leave balance, leave application, and HR policy chat. The app routes user messages to either HR APIs or policy Q&A using a shared intent router.

Architecture

flowchart LR
  UserQuery --> IntentRouter
  IntentRouter --> LeaveBalanceAPI
  IntentRouter --> LeaveApplicationAPI
  IntentRouter --> ChatAPI
  ChatAPI --> HRKnowledgeBase
  ChatAPI --> OptionalLLM
  LeaveBalanceAPI --> MongoOrMemory
  LeaveApplicationAPI --> Response
  HRKnowledgeBase --> Response
  MongoOrMemory --> Response
  Response --> FlutterUI

Loading

Backend

Endpoints:

• GET /health
• GET /metrics
• GET /leave-balance?employee_id=1001
• POST /leave-application
• GET /leave-applications
• POST /chat

See DEMO.md for terminal demo steps, curl commands, sample request/response files, and the Flutter walkthrough.

Set API_KEY to require X-API-Key on HR data and chat endpoints.

Run:

cd hr-backend
npm install
npm test
npm start

With the backend running, use a second terminal:

npm run smoke

The backend uses MongoDB when MONGODB_URI is configured and falls back to seeded in-memory demo data otherwise. GEMINI_API_KEY is optional; without it, policy chat returns deterministic knowledge-base answers.

Docker:

cd hr-backend
cp .env.example .env
docker compose up --build

Kubernetes manifests live in hr-backend/k8s/deployment.yaml and include readiness/liveness probes, resource limits, and Secret-backed configuration. For multi-replica deployments, configure MONGODB_URI so leave applications are stored outside individual application pods.

Dockerfile, Docker Compose, and Kubernetes configuration are validated by static inspection/YAML parsing in this workspace. Runtime container and cluster validation remains a CI or cloud-environment step.

Reviewer Status

• Purpose: Flutter HR assistant with a Node backend for leave workflows and policy chat.
• Quickstart: run backend tests, start npm start in hr-backend, then run npm run smoke.
• Demo path: use DEMO.md for backend curl examples, sample request/response files, and the Flutter walkthrough.
• Deployment status: backend tests and smoke tests pass; Flutter analyze/tests pass; backend Docker/Compose/Kubernetes config is present; backend Docker image builds are validated in CI; cloud deployment is pending.
• Remaining gaps: production HR data integration, identity provider, managed MongoDB, managed secrets, cloud deployment, and policy data governance.
• Portfolio index: https://github.com/Adityansh-Chand/ai-engineering-portfolio

Flutter App

Run:

flutter test
flutter analyze
flutter run -d chrome \
  --dart-define=HR_API_BASE_URL=http://localhost:3000 \
  --dart-define=HR_API_KEY=change-me

Highlights

• BLoC chat flow.
• Shared intent router for tests and production code.
• Backend-hosted chat generation.
• Configurable API base URL via HR_API_BASE_URL.
• Optional frontend API key propagation via HR_API_KEY.
• Backend request IDs, metrics, safe error responses, and leave-application persistence.
• GitHub Actions CI for backend tests/container build and Flutter analyze/tests.
• Local RAG fallback for policy answers.
• Unit tests for routing, model parsing, leave logic, RAG retrieval, and widgets.

License

MIT