Projects
Google Hackathon
Churn Analysis
Project Lead
🏆 3rd Place Winner
This project investigated elevated churn by quantifying lost potential revenue through customer segmentation and targeted analysis across Sales data, usage logs, and support tickets. Our analysis revealed that Enterprise and SMB customers accounted for 65% ($2.5M) of total revenue loss, primarily driven by unmet sales expectations rather than product failure.
Influencer Selection
Advertisement Optimization
Project Lead
This project transformed 400,000 nodes of Twitter network data into a high-impact marketing tool that boosts audience reach by 130% while slashing advertising costs by 75%. Using a custom Streamlit dashboard, businesses can instantly solve complex influencer selection problems by balancing budget, reliability, and demographic targets in real-time. This campaign is formulated as a constrained Maximum Coverage problem in Gurobi
Evision Tesla
Sourcing Strategy
Project Lead
This project involved a comprehensive analysis to determine the optimal location for a major manufacturing facility, focusing on balancing cost savings, risk mitigation, and logistics efficiency across different global locations. The analysis also includes modeling extreme logistics disruptions using 7 simulation-driven decision trees, projected significant cost savings.
BLOCKCHAIN CRYPTOCURRENCY
GYM COIN
Python Model Developer
Gym Coin is a new type of Cryptocurrency we designed to reward exercises and healthy lifestyle. We connected Proof of Work to smart watches like Fit Bit and designed a formula that rewards Gym Coin based on calories burned, exercise time, exercise type etc. There is also a social media component where users can trade Gym Coin for NFTs and interact with other users
MUSIC TIMBRE &
NEWS SENTIMENT ANALYSIS
Python Model Developer
This data visualization project explored the correlation between popular music timbre and news sentiment analysis from 1920 - 2020. We discovered that the Jazz Age (1920-1940) contributed to the strongest positive correlation between negative emotions and Timbre3 (flatness) and strongest negative correlation between negative emotions and Timbre2 (brightness)
JANE STREET MARKET
DATA ANALYSIS
Python Model Developer
This project explored patterns in stock returns and volatility based on a 500 day trading period dataset from Jane Street. We conducted PCA and selected 10 features out of 129 to conduct analysis. We made predictions on number of transactions, returns of specific timeframes, and the inverse relation between volatility and investment time horizon. We also conducted Chi-Square test to evaluate the distribution of trade data.
PATTERNS IN DNA TO PREDICT
ORIGIN OF REPLICATION
Python Model Developer
This project analyzed the patterns of palindromes in CMV virus's DNA to predict the origin of replication so that scientists can take preventative measures to stop the virus from replicating. We conducted Chi-Square tests to compare the distribution of palindromes to Uniform, Poisson and Exponential Distribution using concept of Scan Statistic.
NYPD ALLEGATION PREDICTION PIPELINE
Python Model Developer
This project uses Python Scikit-learn to predict outcome of police allegation based on features of the officer and complainant. This project also uses permutation test to assess the fairness of the prediction model and concluded that the model is accurate and generalizable to different groups of people.
