Team
Jireh
Project Concept
No description has been added yet.
Entry
Status: Submitted
Last saved: May 09 at 5:30 PM IST
Team Roster
Message board not available for this team yet.
Kota Rani Team Lead RSVP Approved
ETL Developer at Accenture
As a solo participant, I was responsible for the end-to-end architecture, development, and deployment of the FormFlow AI engine. My work focused on bridging high-level business logic with dynamic UI generation using the following sponsor tools and APIs:
1. AI Orchestration & Reasoning (Google Gemini)
Integrated Google Gemini 1.5 Pro via the Google AI Studio API to serve as the project's primary reasoning engine.
Designed a custom System Architect Prompt that enables Gemini to translate vague business descriptions (e.g., "I need a retail intake form") into structured UI schemas.
2. Generative UI Protocol (A2UI & CopilotKit)
Implemented the Antigravity (A2UI) protocol to move beyond text-based AI. I leveraged the CopilotKit v2 Runtime to inject the render_a2ui tool into the agent’s loop, allowing it to "push" real-time React components to the user.
Developed the frontend Generative Surface using the @copilotkit/a2ui-renderer, ensuring that the AI-generated forms were interactive and followed business logic constraints.
3. Agentic Backend (LangGraph & Python)
Architected the agentic workflow using LangGraph to maintain state during the multi-step form-building process.
Managed the Python environment using the uv package manager to ensure high-performance execution of the LangChain and Google GenAI libraries.
4. Infrastructure & Integration
Engineered a containerized monorepo environment in GitHub Codespaces, overcoming Windows-to-Linux compatibility hurdles for native AI bindings (esbuild and lightningcss).
Configured the BFF (Backend for Frontend) to handle secure API proxying between the React frontend and the Gemini-powered agent.
Specific Tools Used:
APIs: Google Gemini (Generative AI API), Google Maps JavaScript API (for location-based form fields).
Frameworks: CopilotKit v2, Antigravity ADK, LangGraph, React (Vite).
Infrastructure: GitHub Codespaces
I'm an etl developer more related to data engineering and like to tinker with data and get the output of my desired preference. Currently trying use AI in my current work.
Agentic AI, Deep Learning, RAG
Currently working on an app to generate visuals from a topic to better understand the concepts of any subject using AI.