[DMP 2025]: Text-to-SQL RAG Chatbot for mForm

[ShlokRana]

1) Description

Text-to-SQL RAG Chatbot for mForm is a proposed AI-powered feature designed to enhance mForm — a data collection and management tool for NGOs, CSR initiatives, and philanthropic organizations. This feature will enable users to query their mForm datasets conversationally using natural language. The chatbot will leverage a Text-to-SQL Retrieval-Augmented Generation (RAG) approach, allowing it to understand user intent, translate queries into SQL, retrieve accurate results from the database, and present meaningful insights.

A semantic layer will be included to bridge the gap between technical database schema and user-friendly terminology. This layer will help in mapping human-readable table and column names to their actual technical representations, making the system intuitive and easily configurable for non-technical users.

The goal is to build a production-ready, fully functional solution with a user interface that can be integrated into the existing mForm web application (built on the MEAN stack). The core logic will be developed using Python and NLP techniques.

---

2) Expected Outcome

• A web-based chatbot interface (Angular/MEAN frontend) embedded within mForm.
• The chatbot accepts user queries in natural language.
• Backend (Python-based) converts natural language into SQL using a RAG pipeline.
• Queries run on mForm datasets and return relevant results.
• A semantic layer for mapping friendly labels to database schema.
• Seamless integration with existing mForm authentication and datasets.
• Modular architecture to support future dataset/schema additions.

---

3) Acceptance Criteria

• ✅ Chatbot can be embedded/integrated into existing mForm frontend.
• ✅ Accepts natural language input from users and returns relevant responses.
• ✅ Converts input to valid SQL using a RAG-based approach.
• ✅ SQL runs accurately on mForm datasets and retrieves correct results.
• ✅ Supports multiple mForm project schemas via semantic configuration.
• ✅ Results displayed in tabular or summary form inside the chatbot.
• ✅ All APIs and services should be containerized or easily deployable.
• ✅ Clean, well-documented codebase with setup instructions.
• ✅ Test cases for both frontend and backend components.

---

4) Implementation Details

Frontend:

• Will be developed in Angular and integrated within the existing MEAN stack mForm application.
• A UI component will be created for the chatbot interface, including:
  • Chat bubble
  • Text input field
  • Chat history window
• The frontend will communicate with the Python backend via RESTful APIs to send user queries and display responses.

Backend:

• Will be developed in Python, using either FastAPI or Flask to serve the chatbot backend as a REST API.
• The core logic will include a Text-to-SQL engine built using a Retrieval-Augmented Generation (RAG) approach:
  • Natural language input from users will be parsed and matched with relevant schema context (table/column descriptions, examples).
  • A large language model (e.g., OpenAI GPT APIs) will be used to generate the final SQL query using the retrieved context.
• LangChain or LlamaIndex may be used to build the RAG pipeline effectively.
• A vector database will store the schema context or documentation for retrieval.
• A semantic layer will be implemented to map human-readable terms to the underlying schema:
  • Configuration via YAML or JSON files.
  • Easily extendable and adaptable for multiple mForm project schemas.
• The system will be integrated with mForm’s existing datasets, which have been exported into relational database views for SQL-based querying.

Deployment:

• The backend service will be containerized using Docker for consistent development and deployment.
• A deployment script or guide will be provided for local and production environments.
• The chatbot UI will be modular and pluggable into the existing Angular-based mForm frontend.

---

5) Product Name

mForm

---

6) Organization Name

Dhwani RIS

---

7) Tech Skills Needed

• Full Stack Developer with experience in MEAN Stack (MongoDB, Express, Angular, Node.js)
• Proficient in Python
• Basic knowledge of AI/ML and NLP, particularly in LLMs and Text-to-SQL
• Experience with REST API design, Docker, and frontend-backend integration

---

8) Mentor(s)

@ShlokRana

---

9) Category

Full Stack Development, MEAN, Python, NLP

---

📌 This project is an exciting opportunity to work at the intersection of AI, data, and social impact technology. It will directly empower NGOs and philanthropic initiatives to access insights from their data intuitively.

[Asp-irin]

Hey @ShlokRana ,
I'm interested in taking this up! Maybe we can build an open-source chatbot that other organizations can also use with their own databases. It would also be a great feature for me to add!

To get started, are there any tasks that can help me understand the datasets better — like what the most common queries are and things like that?

Also, could you please share the communication channel where we can discuss ideas?

Thanks!

[aaryan359]

Hi @ShlokRana,

I'm very interested in this project and would love the opportunity to contribute. In the past, I built a chatbot using the Gemini API, which has given me hands-on experience with AI integrations. I believe this background makes me a strong fit for the project.

Looking forward to the possibility of working together!

[MRMORNINGSTAR2233]

Hey @ShlokRana ,
I'm interested in taking this up! because i have already built 90% similar product for a hackathon 1 week before, where I created voice-to-visualization software which is connected to the database and would create a visualization of the data based on the user's voice input as a query.
So i think i am the best person suited for this and I can even convert this into an fullstack application based on your requirements provided. Here is a glimpse of what i built.

git@github.com:MRMORNINGSTAR2233/HTF-I08.git
Thanks!!.

So coming to the implementation part on how i am gonna develop this:

1. Create a interactive Dashboard using Angular along with chat option and also a visualization part if the user wants to visualization of the data
2. Using FastApi to create the API's to connect to the frontend which will also be connected to the database.
3. Using Langchain & RAG along with AI Agents if @ShlokRana agrees , then use these to convert Text-to-SQL and retrieve the related data from the database.(This method will replicate the same methods how humans retrieve data from the DB using SQL commands but here all the work will be done by agents).So there will be automatic mapping of schema and no need to create anything extra.
4. Finally Containerize the backend using Docker and testing it.

[Sharma-Ji-21]

Hi @ShlokRana ,

I’m Om Sharma, a CSE (AI & ML) student with experience in Python, NLP, and full-stack development. I’m really excited about the Text-to-SQL RAG Chatbot for mForm project and would love to contribute.

For implementation, I plan to:
• Build a chatbot UI in Angular
• Use FastAPI + LangChain for a basic RAG-based Text-to-SQL backend
• Add a simple YAML-based semantic layer for schema mapping
• Containerize the backend using Docker

Looking forward to your guidance!

Best,
Om Sharma
sharma-ji-21

[schnrj]

Hey @ShlokRana,

This project sounds super impactful and aligns really well with my interests! I’m a full-stack developer (React, Node.js) with solid experience in Python, REST APIs, and a growing interest in AI/NLP and LLM-based solutions. I've worked on building a chatbot using OpenAI APIs, and I’ve also deployed full-stack apps using MongoDB, Express, and AWS.

I’d love to contribute to the chatbot backend, Text-to-SQL logic, or even help with frontend integration. Happy to jump in wherever help is needed!

You can reach me at b22ee057@iitj.ac.in
GitHub: github.com/schnrj

Thanks!
Sachin Raj

[Alok-Kumar2005]

Hey @ShlokRana
i wanted to work on these project, please assign this one to me

[Raviu123]

Hey @ShlokRana 👋

I’d love to contribute to this project — it's a very familiar concept to me, as I’ve recently built an almost identical project to this ! My project was a RAG-based data visualization and interaction tool using AI, LangChain, and other tools to help users extract insights from their data without writing a single line of code.

This seems like a great fit for my skills and experience, and I’m really excited to help out. Could you please assign this issue to me? 😊

Thanks!

[fizakhan90]

Hey! @ShlokRana

how can someone who is interested in contributing to this project start with it? is there any guidelines for contributors or the contribution will start after the proposal round.

[blenbot]

Hey @ShlokRana
any official communication channel on matrix, slack or discord to get in touch with mentor?

[aryanmahawar205]

Hey @ShlokRana,

I'm Aryan Mahawar, a Computer Science student at VIT Chennai, and I'm particularly looking forward to the Text-to-SQL RAG Chatbot for mForm project under C4GT 2025.

I have thoroughly read the problem statement and just submitted my proposal. I'm especially drawn to this project because it straddles full-stack development and LLM/NLP-based automation, two spaces I'm very passionate about.

This is how I think I can help:

• Familiar with the MEAN stack, and have some experience with Angular and Node.js.
• Proficient in Python, with practical experience in NLP and LLMs, and have recently played around with LangChain and RAG pipelines.
• Experienced with Docker, REST APIs, and semantic layer/config-driven architecture working.
• Have helped open-source libraries such as Optax (DeepMind) and Apache Gravitino to improve clarity and stability.

I'd be happy to contribute to the repo even prior to the selections and get familiar with the existing codebase and priorities. Let me know if there's a decent first issue or any onboarding resources I can jump into!

Thanks, and looking forward to your guidance!

[himudit]

I, Mudit Garg, propose to develop a Text-to-SQL Retrieval-Augmented Generation (RAG) Chatbot for mForm, a tool that supports NGOs, CSR initiatives, and philanthropic organizations with data collection and management. This chatbot will allow users to query mForm datasets conversationally using natural language. By leveraging the power of a Text-to-SQL RAG approach, it will enable the chatbot to understand user intent, convert queries into SQL, retrieve accurate results from the database, and present meaningful insights in an intuitive manner.

The system will include a semantic layer to map human-readable table and column names to their technical counterparts, making it user-friendly and easily configurable for non-technical users. This will allow seamless interaction with mForm datasets while maintaining flexibility for future integrations.

The backend of this solution will be powered by Python and will incorporate a large language model (e.g., OpenAI GPT APIs) to process user input and generate the corresponding SQL queries. I will integrate this solution into the existing mForm web application built on the MEAN stack, ensuring the chatbot fits smoothly within the current system.

By the end of the project, we will have a fully functional chatbot embedded within the mForm application, able to accept natural language input, convert it to SQL, and display relevant data results. This solution will empower NGOs and philanthropic organizations to quickly access insights from their datasets.

I bring experience in full-stack development, Python, and API integration, along with a passion for leveraging AI to drive social impact. My mentor, @ShlokRana, will guide me through this project as we work toward providing intuitive data management tools for organizations in the social sector.

@ShlokRana

[kartiktongaria]

Hey @ShlokRana,

My name is Kartik Raj Tongaria. I have completed my B.Tech in Computer Science and Engineering from NIT Rourkela. I have over 2 years of experience in building and deploying chatbots across various domains.

During this time, I have successfully developed chatbots for multiple companies, catering to diverse business needs such as customer support, automation, and intelligent query handling. My work focuses not just on basic chatbot functionality, but also on enhancing user experience through dynamic learning and adaptability.

One of my key projects includes building a Text-to-SQL chatbot using the Retrieval-Augmented Generation (RAG) framework. In this project, I designed a system where the bot could dynamically train itself by retrieving examples and past interactions, significantly improving its accuracy over time. I also integrated conversation history tracking to maintain context, enabling the bot to provide more precise and relevant responses in follow-up queries.

In addition to this, I have strong expertise in natural language processing (NLP), prompt engineering, retrieval pipelines, and integrating modern AI models like OpenAI, Gemini, and other open-source LLMs into chatbot architectures.

I am passionate about building intelligent systems that continuously learn, adapt, and improve, and I am always eager to explore new advancements in AI, conversational agents, and machine learning.

I have hands-on experience in Python development, API integrations, and working with LLM models. I specialise in creating AI-driven chatbots using NLP techniques and the Retrieval-Augmented Generation (RAG) framework. I am deeply passionate about developing smart, user-centric solutions that can make a positive difference. Guided by @ShlokRana, I look forward to building impactful data management tools for organizations working in the social sector.

[ghostiee-11]

Introduction

I’m Aman Kumar, a full-stack developer and AI enthusiast passionate about building tech that solves real-world problems — especially in areas like education, non-profits, and social impact.

When I came across the mForm Text-to-SQL RAG Chatbot project, I immediately felt connected to its purpose. Giving NGOs and CSR initiatives an intuitive way to interact with their data, instead of digging through complex systems, is exactly the kind of innovation that I want to be a part of.

---

A Bit About Me

• MEAN Stack Experience: Comfortable building scalable applications with MongoDB, Express, Angular, and Node.js.

• Python Backend Development: Proficient with FastAPI, Flask, and creating RESTful APIs.

• AI/NLP Knowledge: Worked with LLMs, LangChain, RAG pipelines, vector databases, and semantic search for previous projects.

• Docker & Deployments: Experienced in containerizing applications and setting up clean CI/CD pipelines.

• Frontend-Backend Integration: Skilled at connecting complex AI backends to user-friendly, responsive UIs.

I’ve worked on conversational AI projects (both voice and text), full-stack search platforms, and data-driven tools — and I would love to bring that energy here.

---

What I’m Excited to Build

Here’s my plan for how I would love to contribute to this project:

1. Frontend Chatbot Interface

• Build a modular Angular component (chat bubble, text input, chat history).

• Integrate cleanly into the existing mForm MEAN stack frontend.

• Enable smooth communication with the Python backend via REST APIs.

2. Backend Text-to-SQL RAG Engine

• Develop a FastAPI (or Flask) service to handle chatbot queries.

• Build a Retrieval-Augmented Generation (RAG) pipeline:

  • Retrieve relevant schema/documentation from a vector database.

  • Use an LLM (e.g., GPT) to generate SQL queries from user input + context.

• Run generated SQL on mForm datasets and retrieve results.

3. Semantic Layer Development

• Build a human-readable mapping between user terms and database schema.

• Set up a flexible, YAML/JSON-based configuration system for easy updates across different mForm projects.

4. Deployment & Documentation

• Containerize the backend using Docker for easy deployment.

• Write clean deployment scripts and setup guides for local and production environments.

• Add test cases for both frontend and backend to ensure robustness.

---

My Approach to the Project

Phase | What I’ll Deliver | Timeline -- | -- | -- Week 1-2 | Build frontend chatbot component, connect APIs | 2 weeks Week 2-3 | Create Text-to-SQL RAG backend pipeline | 1 week Week 3-4 | Implement semantic mapping layer | 1 week Week 4-5 | End-to-end integration, testing with real data | 1 week Week 5-6 | Final polish, documentation, deployment guides | 1 week

---

Why This Project Matters to Me

Data empowers decision-making — but only if people can actually access it intuitively.
Many NGO workers or field managers aren’t tech experts, and making data conversational bridges that gap beautifully.

I genuinely believe that a simple chatbot where someone can ask:

"Show me all beneficiaries enrolled this month in Karnataka"

...and get an answer instantly, can unlock huge impact for organizations doing life-changing work.

This project is the perfect blend of AI innovation, full-stack engineering, and social good — and that’s exactly where I want to be contributing.

---

Working Together

I’m really looking forward to collaborating with @ShlokRana and the Dhwani RIS team.
I’m proactive, take feedback seriously, and believe in delivering clean, maintainable, production-quality work.

[GaurangMundhra]

Hello @ShlokRana,
I'm interested in working on the Text-to-SQL RAG Chatbot for mForm project.

I previously developed a React frontend with an NLP-driven Python backend to answer legal questions autonomously. I had trained the model on the Indian Constitution and secured API key integration along with clean frontend-backend interface.

I think that this experience will enable me to contribute positively to developing the RAG pipeline, semantic layer, and chatbot UI for mForm. I'm confident using NLP, Text-to-SQL, and MEAN stack integration, and I'm looking forward to being able to contribute to making this feature production-ready.

Looking forward to working with you!

[Praveenjhas]

Hey @ShlokRana
I'm really excited about this project and would love to contribute! Just a week ago, I built a similar solution for a hackathon — a voice-to-visualization tool that interprets user voice input, translates it into SQL, and dynamically generates visualizations based on the underlying database. It was a great experience working with NLP, database schemas, and real-time data interpretation.

Given the overlap, I believe I can bring relevant experience and hit the ground running. I’m confident in my ability to turn this into a fully functional full-stack application aligned with your provided requirements. I'm especially excited by the use of a semantic layer and the potential to scale this as an open-source tool for other NGOs.

Would love to know more about how I can get started. Could you please share the datasets or schema documentation to explore typical queries and get a feel for the structure? Also, what's the preferred communication channel for the project team?

Looking forward to collaborating!