Vanshika Nijhawan

Versatile Data Science graduate with hands-on experience in analytics, machine learning, and UX research across technology, ed-tech, and digital media domains. Skilled at transforming user behavior data into actionable insights, designing intuitive product flows, and building ML-driven systems, including emotion recognition, LSTM forecasting, and feature-drift monitoring. Adept at developing dashboards, conducting A/B tests, and delivering client-aligned demos that support data-driven decision-making. Proficient in Python, SQL, SAS, and BI tools, with a proven ability to collaborate cross-functionally and translate complex technical outcomes into clear, high-impact solutions.

1) Automating Video Frame Analysis for Emotion Recognition & Captioning (Python, OpenCV, TensorFlow, CNN–RNN Models)

• Built a real-time video analytics pipeline integrating a CNN-RNN hybrid model for facial emotion recognition and captioning of video frames to identify the emotions of audiences while watching advertisements.
• Engineered frame extraction, preprocessing, and inference workflows using OpenCV and TensorFlow for low-latency performance in live video environments.
• Achieved 82% accuracy on spontaneous expressions by optimizing training pipelines, hyperparameters, and model architecture.
• Designed a reporting framework that converted model outputs into actionable emotional insights for marketing and customer experience strategies.
• The tool could monitor expression distribution, event timelines, and sentiment shifts across live reactions.

2) Segmentation & Sentiment Analysis for Women’s E-commerce Clothing (Python, Pandas, Scikit-learn, NLP, Clustering)

• Conducted customer segmentation using clustering techniques to identify distinct buyer personas and preference patterns across women’s apparel categories.
• Performed sentiment analysis on product reviews, social content, and user feedback using NLP pipelines, enabling data-backed understanding of consumer attitudes.
• Built classification and text-processing pipelines to extract themes influencing purchase behaviour and product satisfaction.
• Synthesized insights into dashboards and summary reports that supported customer-centric product development and personalised marketing campaigns.

3) Stock Market Trend Prediction using LSTM (Python, Keras, Time-Series Modelling, Streamlit UI)

• Developed an LSTM-based deep learning model to forecast stock price trends using historical time-series data, technical indicators, and sequential patterns.
• Engineered preprocessing pipelines for scaling, windowing, and feature extraction to optimise sequence learning and reduce forecasting error.
• Built an interactive dashboard/UI to visualise predictions, actual market movements, and confidence intervals, enabling real-time financial decision support.

• Technical Paper - Automating Video Frame Analysis for Emotion Recognition and Captioning in Real Time (IEEE Publication) - Co-authored an IEEE paper presenting the system architecture, experimental setup, performance evaluation, and reporting framework for my real-time emotion recognition and captioning project.
• Machine Learning with Tree-Based Models in Python (Data Camp Course)
• Recurrent Neural Networks (RNNs) for Language Modelling with Keras (Data Camp Course)
• Linear Classifiers in Python (DataCamp Course)
• Participant in Landmark Forum for Graduates program & “Be Here Now” seminars
• Volunteer at Mumbai Mobile Creche (MMC)

Operating systems: Windows, macOS

Languages: Python, R, HTML, CSS, Base SAS

Databases: SQL, MySQL, PostgreSQL, & MongoDB

Application software: MS Office, Tableau, Power BI, SAS (Visual Analytics, Studio, Enterprise Guide, Enterprise Miner, Viya 4), Looker Studio, GitHub, HDFS

Machine learning: NumPy, Pandas, Matplotlib, SciPy, Scikit-learn, TensorFlow