IAG GBS

Designed and built an AI-powered operational analytics platform for the aviation industry, focused on reducing turnaround times (TAT) and tracking AOG incidents in real time. Led the full-stack development end-to-end, from conversational AI architecture with LLM orchestration to interactive dashboards with advanced data visualization — delivering a system that directly supports operational decision-making at scale.

• Architected and implemented a real-time AI chatbot with SSE streaming, delivering incremental responses, chain-of-thought reasoning steps, and structured data outputs powered by AWS Bedrock (LLM + embeddings).
• Designed conversational AI workflows using LangGraph for multi-step orchestration, enabling context-aware interactions with aviation operational data.
• Built a feedback and traceability system for AI-generated responses, including per-message rating, to continuously improve model outputs and ensure accountability.
• Developed a full analytics dashboard with dynamic charts (Recharts), advanced filtering, and KPI views — enabling coordinators to make data-driven decisions on turnaround and AOG tracking.
• Designed and developed the backend API with FastAPI, delivering ~20 endpoints covering chatbot, analytics, feedback, and operational data use cases.
• Built and maintained the data layer with SQLAlchemy + Alembic, managing 6 schema migrations in PostgreSQL across evolving product requirements.
• Integrated frontend and backend with robust HTTP clients (Ky), JWT authentication, global error handling, and CORS — ensuring security and reliability across environments.
• Applied modern frontend architecture: React 19 + Next.js 15/16 (App Router), TanStack Query, TanStack Table, Zod, React Hook Form, and Radix UI — with clean separation of UI, logic, and data layers.
• Implemented comprehensive testing: unit tests with Vitest, E2E tests with Playwright (multi-browser), and a backend suite of 47 test files covering APIs, services, middleware, and AI workflows. Documented components with Storybook.
• Enabled production observability with Datadog/ddtrace for monitoring, diagnostics, and distributed tracing across frontend and backend services.