Introduction

Publications and Presentations

Ahmed KS, Faisal R, Ali MM, Virani SS, Marcinak CT, Zafar SN. Predicting Post-discharge Infection: A Machine Learning-driven Composite Outcome Model for Patients Undergoing Pancreatectomy. Accepted for e-poster presentation at the American College of Surgeons’ Clinical Congress 2025.
Faisal R, Yan F, Omar C. From Holes To Whole: Building A Guided Onboarding Experience for Functional Programming in Hazel. Poster presentation at the University of Michigan Explore Computer Science Research Poster Session 2025.

Conducted comprehensive analysis of the Global Weather Repository dataset containing over 46,000 rows of meteorological data, including temperature, precipitation, and air quality indicators.
Designed and implemented advanced anomaly detection techniques (Z-Score, IQR, Isolation Forest) to identify statistical and environmental outliers.
Developed and compared time-series forecasting models (SARIMA, Random Forest, Stacked Ensemble) to predict temperature and precipitation trends with a focus on anomaly scenarios.
Explored geographical and spatial analysis through choropleth maps, highlighting regional variations in weather conditions and anomalies.
Evaluated feature importance using SHAP values, identifying wind speed and humidity as key drivers of air quality.

Technologies: Python, Scikit-learn, Pandas, NumPy, Plotly, Graphviz

Built a scalable search engine similar to Google or Bing, for EECS 485.
Includes information retrieval concepts like text analysis (tf-idf) and link analysis (PageRank), and parallel data processing with MapReduce.
Uses a Service-Oriented Architecture to scale dynamic pages and web search.
Creates a segmented inverted index of web pages using a pipeline of MapReduce programs.
Built an Index server, a REST API app that returns search results in JSON format.
Built a Search server, a user interface that returns search results just like Google or Bing.

Technologies: Python, SQL, Flask, HTML and CSS, JSON

This project leverages logistic regression to measure the predictive performance of Case Detection Rate (CDR) in country-specific mortality outcomes for patients with a dual burden of Tuberculosis (TB) and Human Immunodeficiency Viruses (HIV).
Using data from the Tuberculosis Burden by Country dataset, it evaluates the role of CDR in determining mortality-to-incidence ratios (MIRs).
Feature engineering techniques such as logarithmic scaling and quantile transformation were applied to the baseline model to address data skewness and improve model performance.
The final model demonstrated an improvement in AUC for dual-burden mortality predictions, emphasizing the importance of early case detection in mitigating public health challenges.

Technologies: Python, Scikit-learn, Pandas, NumPy, Plotly

MapReduce framework in Python inspired by Google’s original MapReduce paper for EECS 485.
Executes MapReduce programs with distributed processing on a cluster of computers like AWS EMR, Google Dataproc, or Microsoft MapReduce.
Includes program execution, distributed systems, fault tolerance, OS-provided concurrency facilities (threads and processes), and networking (sockets).
Consists of a Manager which listens for user-submitted MapReduce jobs and distributes the work among Workers, and multiple Worker instances that receive instructions from the Manager and execute map and reduce tasks that combine to form a MapReduce program.

Technologies: Python, Madoop (Custom version of Hadoop)

Developed an Instagram clone as part of a three-project sequence for EECS 485, focusing on building client-side dynamic pages using JavaScript, React, and AJAX.
Refactored the server-side logic from Flask into a REST API to handle asynchronous data requests, enabling real-time updates for user interactions such as likes, comments, and infinite scrolling without page reloads.

Technologies: Python, SQL, HTML, CSS, JavaScript, React, Flask, AWS