Aug 2022 – Jul 2025 · 2 yrs 11 mos · India · Remote

• Project 1
• Developed and implemented a Theme Model leveraging NLP to analyze and extract themes and sentiments from employee' texts.
• 1. Applied Ensemble Models for Sentiment Analysis.
• 2. Topic Modelling using LDA.
• 3. Implemented a Zero-Shot Classification model from Hugging Face to classify text data.
• 4. Employed GPT-4o-mini, along with prompt engineering techniques to confine classifications to bounded classes tailored to specific categories, enhancing Zero-Shot Classification.
• 5. Applied Fine-tuning to further refine model performance and achieve a 98% accuracy rate.
• Project 2
• Built a Disengagement Model for multiple companies to categorize their employees into appropriate groups in order to understand employees’ disengagement in their organization and be able to reduce the churn rate.
• 1. Optimized and created employee segmentation using RFMBC models, aimed at improving user engagement and reducing churn rate.
• 2. Implemented Feature Engineering, Variable Combination and Classification.
• 3. Used Statistical Techniques WOE and IV to evaluate the predictive power of the variables and for creating a Variable Combination.
• 4. Collaborated with HR teams to integrate the Model into existing processes, allowing for early identification of at-risk employees.
• 5. Successfully reduced attrition rates, based on insights generated by the Model. Improved user engagement by 65% and reduction in churn rate by 25% within six months.
• Project 3
• RAG-based Chatbot API for Policy Document Q&A
• 1. Built a Flask-based API to serve a Retrieval-Augmented Generation (RAG) pipeline for querying policy documents from multiple organizations.
• 2. Preprocessed documents, generated vector embeddings using OpenAI’s model, and stored FAISS indexes locally
• for fast semantic retrieval.
• 3. Designed dynamic index loading and similarity search based on organization-specific context.
• 4. Integrated GPT API to generate accurate, context-aware answers from top-k retrieved content.

Dec 2019 – Jul 2022 · 2 yrs 7 mos · Hyderabad, Telangana, India

• Machine Learning Projects:
• 1. Fuel consumption rate Analysis in Python
• Feature engineering has been done using PCA to reduce the number of predictor variables
• Data pre-processing involved missing value imputation and Outlier detection for each of the variables
• Applied Multiple Regression technique involving multiple parameters to predict C-rate
• Improved the accuracy of the model using Random forest and Boosting technique with around 92% accuracy
• EMS/Non-EMS Fuel savings
• Analyzed each of the routes independently to pre-process the data
• Random forest technique is used to choose the important features for each of the routes and to get the fuel prediction
• Computation of fuel savings for EMS trains using trained model of Non-EMS
• 2. Prediction of subscription of Term Deposits for bank clients in Python using Machine Learning
• Data pre-processing involved Missing value imputation and Outlier detection for each of the variables.
• Feature engineering has been done using Standardization & Handled Categorical Features using One Hot Encoding.
• Applied SMOTE (Synthetic Minority Oversampling Technique) to handle Imbalanced Dataset.
• Applied Logistic Regression technique.
• Applied Recursive Feature Ellimination to repeatedly construct the Model and choose either the best or worst performing feature.
• Computation of Accuracy of the model using Confusion Matrix and ROC.