statathon nova project about survey design using ai

project

1. STATATHON 2025 Problem Statement ID: 03 Problem Statement Title: AI-Powered Smart Survey Tool for Survey Data Collection PS Category- Software/Hardware: Software Team ID: 6677 Team Name (Registered on portal):GUNAL S @ STATATHON Idea submission- Template

2. Questero AI - AI-Powered, Multilingual and Real Time Data Collection app Questero AI is an offline-first,multilingual survey app with AI Enumerator Co-pilot that can be listen,adapts the question and validte surveys in offline(local storage) Our solution transforms 40% faster , 50% fewer errors and 100% inclusion of across all languages and regions across india with reduced training time of Enumerators upto 2-3 days. Questero AI is the first survey platform to bring offline-first,Multilingual and Ai-assistance into Government-scale socio-economic Data Collection . @ STATATHON Idea submission- Template GUNAL S

3. TECHNICAL APPROACH React Native – Offline-first multilingual mobile app. React + TailwindCSS – Web dashboard for survey creation & monitoring. FastAPI + PostgreSQL – Backend APIs & structured data storage. Redis – Task queues & offline sync handling. Whisper – Speech-to-text for Hindi & regional languages. Google Translate API – Language translation. LLM + Rules Engine – Adaptive questioning & AI Enumerator Co-Pilot. Transformers – Auto-coding Twilio Voice – Multi-channel survey delivery (chatbot). Autoencoder models -anomaly detection @ STATATHON I dea submission- Template GUNAL S Work flow Daigram

4. FEASIBILITY AND VIABILITY Analysis of the feasibility Building on proven technologies like react native,fastAPI,whispor, Questero AI offeres offline capability,integration of AI Enumerator Co-pilot can be technilcally achievable. Potential challenges Speech-to-text models may give Accent related transcription error affecting AI Co-Pilot guidance in surveying. Strategies to overcome challenge Fine Tunning (Whispor) with regional indian speech dataset for maintain accuracy @ STATATHON Idea submission- Template GUNAL S

5. IMPACT AND BENEFITS Potential Impact : Inclusive Reach,Used across multiple regions of India , Faster DataCollection , Improved Data Quality , Empowered Field Workplace. Benefits: Social Benefits : Our solution enables participation from diverse linguistic and social group across india Economic Benefits : Reduces survey completion time by 40% Public Benefit s: Provides timely-quality data enabling faster response Cost Effectiveness : Automates time-consuming tasks , Reducing Expense in Data collection @ STATATHON I dea submission- Template GUNAL S

6. RESEARCH AND REFERENCES ASR Model Performance on Indian Languages (2025) Comparative performance analysis of end-to-end ASR models on Indo-Aryan and Dravidian languages link:https://asmp-eurasipjournals.springeropen.com/articles/10.1186/s13636-025-00395-5 @ STATATHON Idea submission- Template GUNAL S Improving Survey data quality using LLms We have closely examined how survey data quality are improved using llms,monitoring real-time data in survey link : https://www.geopoll.com/blog/llms-improving-survey-data-quality IndicSUPERB : A 1,684 hours speech dataset across 12 indian langauges , supporting ASR, lnguage ID , Keyword spotting link :https://arxiv.org/abs/2107.07402 other references: Telephone Surveys Meet Conversational AI: https://arxiv.org/abs/2502.20140