Real Estate Price Prediction & Trend Analysis
Summary:
Developed and deployed a machine learning model to predict property prices with high accuracy, enabling better negotiation strategies for the sales team.
Problem:
The sales team needed accurate and up-to-date property valuations to negotiate resale deals effectively.
Approach:
-
Collected and processed structured and unstructured real estate transaction data.
-
Performed EDA, feature engineering, and model training using ML/DL algorithms.
-
Achieved 90% accuracy in predicting property prices.
-
Integrated trend analysis for forecasting market movements.
-
Deployed the model via Flask API on AWS EC2 for real-time use by the sales team.
Tech Stack: Python, Pandas, Scikit-learn, AWS EC2, Flask, Power BI
AI Agent for Real Estate Data Crawling
Summary:
Built an autonomous AI agent that intelligently crawls property listings and market data from multiple sources, cleans it, and stores it for analytics and ML model training.
Problem:
The company relied on costly third-party datasets and manual research, which slowed down market analysis and increased operational costs.
Approach:
-
Designed an AI agent using LangChain to decide when and what property data to scrape.
-
Integrated BeautifulSoup and Selenium for structured and dynamic data crawling.
-
Used n8n to automate crawling of static websites.
-
Automated scheduling for daily crawls and incremental updates.
Tech Stack: LangChain, Python, BeautifulSoup, Selenium, Requests, Pandas,n8n
AI Agents for Property Document Search & Recommendations
Summary:
Built custom AI agents using LangChain and Retrieval-Augmented Generation (RAG) to automate property recommendations and document retrieval for internal teams.
Problem:
Searching through vast property documents and datasets was time-consuming for operations and sales teams.
Approach:
-
Indexed property documents using vector databases for semantic search.
-
Integrated LLM-powered AI agents to answer property-related queries.
-
Automated recommendations based on user preferences and past transactions.
Tech Stack: LangChain, OpenAI API, FAISS, Python
Automated Content Writing, Summarization, and Document Intelligence for Marketing
Summary:
Developed AI-powered solutions for the marketing team to generate property descriptions, summarize market reports, and extract key details from property documents using NER and OCR.
Approach:
-
Fine-tuned LLM prompts for high-quality, SEO-optimized property descriptions in multiple languages.
-
Implemented automatic summarization for lengthy market reports and property listings.
-
Applied Named Entity Recognition (NER) to extract critical details like property names, locations, builder names, and RERA IDs from unstructured descriptions.
-
Integrated OCR pipelines to digitize scanned property documents, enabling automated extraction of structured data.
Tech Stack: OpenAI API, LangChain, Python, Mistral
Lead Scoring & Conversion Prediction
Summary:
Created a classification-based ML model to score CRM leads and predict their likelihood of conversion.
Approach:
-
Preprocessed CRM data, handled missing values, and engineered lead-specific features.
-
Implemented Bagging & Boosting algorithms.
-
Achieved 85% accuracy in predicting high-value leads.
Tech Stack: Python, Scikit-learn, XGBoost, Pandas
Real-Time Loan Performance Dashboard
Summary:
Built an interactive Power BI dashboard for monitoring the loan team’s performance and the real estate market.
Approach:
-
Designed a data warehouse using Fact and Dimension tables.
-
Automated ETL pipelines via Power Query.
-
Integrated predictive analytics for loan performance trends.
Tech Stack: Power BI, SQL, DAX, Power Query
Property Price Prediction
Building a linear regression model in R, Python we predict the final price of a property on the basis of its feature including no of room, area ,bathroom, wiring, interior etc.
Churn analysis telecom industry
Using classification model logistic regression, Decision Tree we build a model on the basis of previouscustomer data to predict whether the customer will churn or not .
Bank Credit Card Default
Using classification algorithms KNN, Logistic Regression, Decision tree , we build a model which predict about the customer who is applying for credit card will do default or not
Bank Loan Data
Using a linear regression model it has to predict that how much percent of interest for the loan should be given to customer, on the basis of fico-range and cibil score.
.jpg)
Car Sale Data
Heart attack analysis & Prediction
Using 10 classification algorithms , we build a model on the basis of the variable eg. age, gender, cholesterol, BMI etc. to predict the probability of a patient to get heart attack
Using linear regression model we find the key factor eg.mileage,weight, no of cylinders, stroke etc. which drives the price of car