Heuristics Pharma Perception – Data Analyst

• Engineered a model-serving infrastructure using Kubernetes and FastAPI, reducing model inference time by 30% while enabling scalable microservices for model APIs.
• Developed an embeddings-based similarity model using FAISS and OpenAI embeddings, enhancing content-based retrieval and improving response accuracy by 15%.
• Built and fine-tuned NLP pipelines leveraging BERT models for automated entity recognition and classification in medical reports, resulting in 20% increase in processing speed.
• Integrated a high-speed vector search pipeline with Neo4j and Elasticsearch to enable semantic question-answering over large datasets with latency under 300ms.
• Automated model monitoring and retraining workflows using MLflow & Prefect, achieving continuous model improvement and reducing manual intervention by 70%.
• Orchestrated ETL workflows for unstructured data processing using Apache Airflow and deployed data transformation scripts across multiple data lakes and APIs.

Sriroz Consultants Private Limited – Business Analyst

• Redesigned SQL-based sales reports to integrate real-time customer purchase behavior analysis, reducing report generation time by 30% and enabling faster decision-making.
• Performed exploratory data analysis (EDA) on customer retention patterns using Python and MySQL, identifying key churn factors and suggesting strategies that reduced churn by 12%.
• Developed interactive Power BI dashboards that automated sales reporting and improved forecasting accuracy by 25% through data modeling and DAX optimization.
• Integrated Python-based CRM automation using Pandas and Selenium to track lead-to-conversion journeys, increasing conversion by 20% and adding $105K in revenue.
• Conducted competitive pricing analysis using web-scraped competitor data and market trends, which informed dynamic pricing strategies that boosted profit margins by 12%.
• Implemented customer segmentation models (K-means and decision trees) that improved targeted marketing campaigns, leading to a 22% increase in repeat purchases.

AI-Powered Patient Risk Prediction System

• Developed a machine learning model for early prediction of patient readmission risk using XGBoost & Random Forest algorithms.
• Automated patient data ingestion from EHRs and wearables using Apache Airflow and stored in Snowflake warehouse.
• Built and monitored live pipelines with Apache Spark Streaming and Kafka, enabling real-time risk dashboarding.
• Implemented BERT models to extract risk indicators from unstructured physician notes, improving feature richness by 30%.
• Deployed containerized model to HIPAA-compliant Kubernetes infrastructure on AWS, ensuring compliance and scalability.

Emotional Chatbot for Multimodal Emotion Recognition

• Built emotion recognition chatbot integrating text-based NLP (Random Forest) and voice-based CNN model (Keras) with 90% accuracy.
• Designed modular structure supporting APIs for music (Spotify), videos (YouTube), activities (Eventbrite), and podcasts (Listen Notes).
• Trained models on augmented datasets with oversampling and feature engineering for improved F1-score of 0.88.
• Developed and deployed UI using Flask and Bootstrap for real-time usage in mental health and education domains.

Pace University, Seidenberg School of Computer Science and Information Systems

MS in Information Systems | Concentration: Information Systems | Honors: Masters

Savitribai Phule Pune University – Modern Institute of Business Studies

MBA | Concentration: Business Analytics | Honors: Masters

Savitribai Phule Pune University – Dr. D. Y. Patil College

BBA | Concentration: HR Management | Honors: Bachelors