Past Engagements 

Tech Stack: Python, Pandas, NumPy, Scikit-learn, XGBoost, LightGBM, BioBERT, Streamlit

 Built an end-to-end predictive analytics solution using structured clinical trial metadata and NLP-based text features.

 Performed data cleaning, preprocessing, feature engineering, dimensionality reduction, and class imbalance handling to improve model performance.

 Developed and evaluated Logistic Regression, XGBoost, and LightGBM models using Precision, Recall, F1-score, and ROC-AUC metrics.

 Implemented SHAP-based model explainability and developed a Streamlit dashboard for interactive prediction visualisation.

Tech Stack: Apache Kafka, Spark, Hive, HDFS, MongoDB, Presto, Grafana

 Designed a scalable big data architecture for processing high-volume digital marketing campaign data in real time.

 Proposed data ingestion, storage, processing, and visualisation pipelines using Apache Kafka, Spark, HDFS, and cloud-based components.

 Designed a hybrid architecture supporting both real-time reporting and batch analytics.

 Developed an analytics workflow for campaign reporting, dashboard generation, and historical trend analysis.

Tech Stack: Python, TensorFlow, OpenCV, CNN, LSTM. Developed a deep learning framework to detect suspicious highway activities using surveillance video data.

 Applied image preprocessing, feature extraction, and temporal sequence modelling for anomaly detection.

 Evaluated model performance using Precision, Recall, and F1-score for real-time deployment.

 Designed the solution to support intelligent traffic monitoring and automated incident detection.