Find the Best
Electronics Deal
Instantly.
Compare prices across Amazon, Flipkart, Best Buy, eBay and Google Shopping. Ranked by our AI-trained Buy Score algorithm.
Built with AI. Backed by Data.
ElectroFind is a final-year engineering project combining machine learning, real-time web scraping, and modern frontend to solve the electronics price comparison problem.
What we built
A full-stack AI-powered electronics comparison platform. We collected, cleaned and vectorized product data from Amazon and Flipkart, trained a similarity model, and built real-time search that surfaces the best deal across 5 major platforms — ranked by a custom Buy Score.
Data Collection
Scraped 50,000+ electronics listings from Amazon and Flipkart using automated pipelines.
AI Similarity Engine
Cosine similarity and TF-IDF vectorization to match and deduplicate products across platforms.
Data Processing
Multi-stage ETL pipeline: cleaning, normalization, feature engineering and enrichment.
Buy Score Algorithm
Custom scoring model combining price, rating, and review signals into a single 0–100 score.
Real-Time Search
Live product fetching from 5 platforms with sub-second aggregation and ranking.
Full-Stack Architecture
Next.js 14 frontend, Node.js API layer, Python ML backend, and REST integration.
50K+
Products Collected
94%
Match Accuracy
5
Live Platforms
From idea to production
Project Inception
Defined scope, formed team, identified data sources — Amazon and Flipkart electronics categories.
Data Collection Pipeline
Built automated scrapers to collect 50,000+ product listings. Structured raw data into CSV datasets.
Data Cleaning & EDA
Removed duplicates, handled missing values, normalized price formats, performed exploratory analysis.
Feature Engineering
Extracted title embeddings, category tags, brand signals, and price-tier features for model input.
Model Training
Trained cosine similarity and TF-IDF vectorization models. Validated deduplication accuracy at 94%.
Backend Development
Built REST API layer integrating Oxylabs and Serper APIs. Implemented scoring and aggregation logic.
Frontend Development
Designed and built Next.js frontend with Tailwind CSS, Framer Motion animations and ShadCN components.
Integration & Testing
End-to-end integration testing, performance benchmarking, and cross-platform result validation.
Beta Launch
Deployed beta version. Collected user feedback, iterated on UI/UX and scoring algorithm.
Production Release
Full production deployment with 5 platform support, real-time search, and AI Buy Score.
Who built ElectroFind
A team of three AI engineering students bringing together machine learning, data engineering and full-stack development.
All team members — B.E. Artificial Intelligence · Final Year Project 2025–26
What's coming next
ElectroFind is actively evolving. Here's what the team is building toward in the next development phases.
Deeper AI Recommendations
Fine-tune the similarity model with user interaction data to provide personalized product recommendations based on browsing patterns.
Expand to 10+ Platforms
Add support for Walmart, Newegg, Croma, Reliance Digital, B&H Photo and international storefronts for broader coverage.
Price Drop Alerts
Let users set price thresholds on any product. Get notified via email or push notification when the price hits their target.
Native Mobile App
React Native application for iOS and Android with barcode scanning to instantly find the best online price for in-store products.
Price History Charts
Track and visualize price changes over time using our collected historical dataset and real-time price monitoring pipeline.
Verified Reviews Layer
Cross-reference reviews across platforms using NLP to detect fake reviews and surface only verified, trustworthy ratings.
Have a feature idea? Reach out to the team →