Shambhavi Chati

Mobile: 312-***-**** Email: adlbw8@r.postjobfree.com https://www.linkedin.com/in/shambhavichati/ EDUCATION

Master of Science in Business Analytics Aug 2019 – December 2020 University of Illinois at Chicago

B. Tech in Electrical & Electronics Engineering Jun 2014 – May 2018 Vellore Institute of Technology, Vellore, India

SKILLS

Programming Languages: R (Tidy Verse, ggplot), Python (Pandas, NumPy, Scikit-Learn, Spark, Keras, TensorFlow PySpark, PYOD), HTML, CSS, Flask.

Database and Servers: Elastic Search, MySQL, and Hadoop. Visualizations: Tableau, PowerBI, Google Analytics, Seaborn, RapidMiner, Scipy. ML: Random Forest, XGBoost, Naïve Bayes, Logistic Regression, RNN, CNN, LDA, Lasso Regression, Clustering (k- NN, K-means).

Technologies: AWS-EC2, S3, SageMaker, Azure, GCP, Hadoop, Jira. EXPERIENCE

ThinkRisk.ai, New Jersey, USA September 2020 – Present Data Scientist (Fraud and Risk)

● Developed and optimized unsupervised ML models (isolation forest and autoencoders) for financial fraud and anomaly detection, product development using Python, TensorFlow and PYOD.

● Monitored $30,000 fraudulent activity over two months by developing a risk assessment algorithm and improved the risk identification in financial ledgers.

● Deploying fraud detection machine learning models with large relational databases using python scripts on Microsoft Azure cloud for batch-based training and testing increasing performance by 10% using CI/CD. LabelMaster, Chicago, IL, USA ) September 2020 – December 2020 Data Scientist

• Developed a Vector Auto Regression and LSTM based models to predict department wise future sales saving

$150,000 across three departments.

• Extracted multiple time series features for predictive data modelling to capture the impact of COVID-19 on sales and econometric factors reducing manual forecasting effort by 40%.

• Developed ETL pipelines to gather sales and economic indicator data from client’s data sourcing platform using EC2, MongoDB Python and shell scripting using best practices, making the process 30% faster.

• Deployed a Flask application for the client using Amazon Web Services. Vindiata Consulting, Mumbai

Data Scientist June 2018 – Aug 2019

● Fraud : Designed an ML-based online gaming fraud detection product. Designed back-end architecture and ETL script. Created a Random Forest based algorithm to predict fraudulent users on the platform with an accuracy of 89%, saving $30,000 over 3 months of fraudulent cashouts.

● Prepared a collection of fraud data for risk identification by extracting, cleaning, and merging from different tables using MySQL, increasing efficiency of marketing analysis by 15%.

● Cleaned over 700,000 records of player profile data for the past 10 years and performed Exploratory Data Analysis using Tableau and Python.

● Marketing Promotions: Increased platform traffic by 25% by creating K-means based targeted promotions for different players on the platform. Used A/B tests to evaluate performance. PROJECTS

MITRE Competition 2019

• Designed a complete data cleaning pipeline for handling data from diverse sources. Our LSTM based model achieved 91.4% accuracy. Current work is focused on developing to help label more data and classify fraudulent activities and transactions in healthcare and pharma. Natural Language Processing: Extract Stock Sentiment from News Headlines

• Performed web scraping using Selenium and Beautiful Soup to obtain news headlines about AMZN and TSLA stocks. Applied sentiment analysis to predict whether the market feels good or bad about a stock, post COVID, precision 86%.

MBTI prediction full stack app

• Built a reproducible and modular pipeline with a Flask app to predict MBTI personality type using the XGBoost model hosted in a Docker container and stored user inputs and predictions in RDS MySQL database.

Contact this candidate