SAI CHARAN B

My interest in business analytics led me to pursue a PGDBA from IIM Calcutta, IIT Kharagpur, and ISI Kolkata, where I gained extensive knowledge of machine learning, statistics, deep learning, and their practical applications in the business realm. Throughout my academic journey, I successfully executed multiple real-time data science projects that delivered tangible business outcomes, fostering the development of a robust skill set. I worked as a data science intern at Airtel Digital. Currently, I'm working as a Senior Business Analyst at Tech Mahindra I am actively seeking new opportunities as a data scientist to further my career growth.

• Certified in the 'Green Belt Programme' on Lean Six Sigma Methodology from KPMG India (13th Feb 2022)
• Certified in 'Winter School on Deep Learning' from perceptron to transformers offered by ISI Kolkata'22
• Currently pursuing certification in 'Azure dp-203' data engineering on Microsoft Azure by Koenig-solutions.

• National Finalist in ISB Datathon, a retail store clustering and sales prediction analytics case competition '21
• National Finalist (top 10) in 'Shri Ram Analytica', a case study on education upskilling hosted by SRCC'22
• National 2nd runner-up in Cryptx Competition, an analytical case competition organized by Spectra IIMB '22

Athletics and Other Sports & PORs Won Gold medal in 3km and 1.5km run in the under-16 category in national level CBSE Cluster 7 athletic meet Won 2nd position in Tennis Competition during the Intersectional 2022, conducted by IIM Calcutta

• Late Delivery prediction: Developed a model using 180k+ transactions to predict late deliveries. Handled class imbalance with SMOTE, evaluated with k-fold CV, and achieved a max F1-score of 0.81 after hyperparameter tuning with Grid search. Web App Stock
• Prediction (Time Series): Web Scraped data from the Yahoo Finance site, and created a dashboard on Streamlit. Diagnosed stationarity using ADF & KPSS test; performed STL decomposition & Box-Cox transformation, identified ARIMA order with ACF, PACF. Improved RMSE by 30% with LSTM (0.2 dropout) compared to ARIMA and Auto-Arima.
• Sentiment Analysis (Using NLP, DL): Predicted the rise/fall of the DJIA using the top 25 news headlines of the day. Performed text processing with NLTK, created word embeddings using TF-IDF, and CBOW, and compared the results. Utilized Logistic Regression, BiLSTM with Dropout models via Tensorflow and improved the F1-score to 0.85.
• Car Price Prediction: Built linear & non-linear regression models to predict second-hand car prices using 8k+ records. Checked multicollinearity with VIF & analyzed data with KS-test. Used Lasso, Ridge & Random Forest regression models to improve accuracy. Random Forest reduced RMSE by 42%.
• Recommendation System: Created a recommendation system to suggest practice problems by analyzing data from 3500 users and 6500 problems. The system utilized Collaborative Filtering to recommend problems that are commonly solved by nearest neighbors.
• Customer segmentation: Performed customer segmentation to run marketing campaigns using online retail transaction data. RFM analysis was carried out to segment customers, and K-means clustering was used to obtain the customer segments. The elbow method was utilized to determine the optimal number of segments, resulting in a silhouette score of 0.595.