Differentiable optical lens simulator for end-to-end computational imaging.

Developed an end-to-end ML system on Azure to predict loan defaults, leveraging advanced data preprocessing, feature engineering, and machine learning models to optimize accuracy. This project includes a comprehensive suite of tools and techniques for robust financial risk assessment, deployed to enhance decision-making for high-risk exposures.

[AAAI 2022] Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics.

End-to-end PSF extraction for 3D microscopy

Deep Negative Volume Segmentation - automated 3D CT segmentation of body joints for dentistry

Supported Models: MobileNet [V1, V2, V3_Small, V3_Large] (Both 1D and 2D versions with DEMO, for Classification and Regression)

A Fully Automatic Method for Predicting Contact Maps of RNAs by Evolutionary Coupling Analysis

End-to-End ETL Pipeline for Film Data Crawling from Ohitv

End-to-end data engineering pipeline with real-time streaming, cloud processing, and analytics. Built with Apache Kafka, Spark, AWS Glue, and Snowflake using Apache Iceberg tables.

CI/CD pipeline to deploy Node.js app to AWS using Terraform, Ansible, and GitHub Actions. Fully automated deployment on every push to main branch

End-to-end Fraud Detection MLOps pipeline integrating MLflow, FastAPI, Streamlit, Docker, Kubernetes, Prometheus, and Grafana for real-time fraud prediction, experiment tracking, and monitoring.

This project uses machine learning to predict customer churn in the banking sector. It covers the end-to-end process, from data ingestion, validation, and transformation to model training and deployment using FastAPI. The system includes real-time predictions and provides an API for customer churn analysis.

Analyzed a multicategory e-commerce store using big data techniques on a Kaggle dataset with the help of AWS EC2, AWS S3, PySpark, AWS Glue ETL, AWS Athena, AWS CloudFormation, AWS Lambda and Power BI!

⚡ E-commerce Data Engine Processing 10,000+ records with a custom Python pipeline. Includes advanced data imputation and interactive Business KPI dashboards. 📈

Detect U.S. housing market bubbles using macroeconomic signals. Forecast HPI, score speculative risk, and visualize insights using a fully modular, cloud-native GCP pipeline.

This project is an end-to-end MLOps pipeline for a network security system that detects phishing and malicious activities using machine learning. It automates data ingestion, preprocessing, model training, and deployment while leveraging AWS S3 for model storage and GitHub Actions for CI/CD. The system includes realtime monitoring & a web interface

Successfully established a machine learning model which can accurately predict the expected life duration of a human being based on several demographic features such as alcohol consumption per capita, average BMI of entire population, etc.

This project is a NLP problem that will be the foundation of an English program used by the company Easy Sailing Language Training.

A production-grade end-to-end machine learning pipeline for predicting student performance, featuring automated project setup, data ingestion from PostgreSQL, modular preprocessing and model training, and experiment tracking using MLflow and Dagshub.

Full-stack MLOps pipeline for predicting colorectal cancer patient survival using Gradient Boosting, Kubeflow Pipelines, MLflow, and Flask. Designed for hospitals, researchers, and real-world healthcare applications.