Data Scientist

TANIYA RAJANI Data Scientist

linkedin.com/in/taniyarajani github.com/taniyarajani *******@***.*** 773-***-**** Chicago, IL EDUCATION

University of Illinois, Chicago: Master’s in Business Analytics, GPA:3.9/4 Aug 2019 - Dec 2020 SGSITS, India: Bachelor’s in Industrial & Production Engineering, GPA:3.8/4 Aug 2012 - May 2016 MACHINE LEARNING PUBLICATIONS

M. Tanveer, T. Rajani, M.A. Ganaie (’19). Improved sparse pinball twin SVM. IEEE International Conference Systems, Man and Cybernetics (SMC).

M. Tanveer, T. Rajani, R. Rastogi, Y.H. Shao. (’19). Comprehensive review on twin support vector machines. ACM Computing Surveys (CSUR)**

. [SCI Indexed with Impact Factor: 6.13]

TECHNICAL SKILLS

Tools: Python, PyTorch, Tensorflow, Keras, scikit-learn, R, C, MATLAB, Tableau, SQL, Hive, Spark, AWS, EC2, GIT Techniques: GLM, KNN, PCA, XGBoost, RF, Ensemble, Clustering, LDA, SVM, NN, CNN, RNN, NLP, A/B Testing WORK EXPERIENCE

ClarisHealth*

, IL, USA Data Scientist Intern May 2020 - Present

Unsupervised Healthcare Provider Fraud Detection

- Ideated various features using medical claims data in PostgreSQL for peer-to-peer analysis of providers

- Designed deep LSTM Autoencoder model to generate provider anomaly score; achieved 0.67 specificity University of Illinois, IL, USA Graduate Assistant, Data Science Jan 2020 - May 2020

Unsupervised Network Analysis to Identify Medical Episodes of Care

- Built a directed network graph of procedure codes and defined clusters of medical episodes of care using single and mixed-membership clustering

- Studied various parameters, defined metrics, and collaborated with subject matter expert for validation Indian Institute of Technology (IIT), India Research Associate, Machine Learning Jan 2019 - Jul 2019

Research on Twin Support Vector Machines (Guide: Dr. M. Tanveer)

- Formulated a novel variant of twin SVM; achieved 4.93% higher accuracy than other baseline SVM variants Bira91, India Assistant Manager, Analytics Apr 2018 - Aug 2018

Modelling & Data Visualization

- Build a GBM response model to increase conversion rate of beer outlets; model achieved 78% accuracy & 0.82 Gini on out of sample data; resulted in potential $15k monthly savings

- Developed interactive visualization using India beer sales data on Tableau to generate actionable insights

- Automated data extraction from AccuWeather API using selenium in Python & eliminated manual efforts Godrej Consumer Products, India Senior Executive, Analytics Jun 2016 - Nov 2017

Operations Analytics

- Created an OCR engine to extract structured tables from image contract documents in Python; $50K savings

- Spearheaded planning for a plant & enhanced productivity of a product by 5%; detected key bottleneck PROJECTS

- Leveraged pre-trained CNN encoder & LSTM as decoder for Image Captioning on Flickr8k; 0.65 BLEU score

- Built a sequence to sequence LSTM model with attention for Neural Machine Translation task; 0.60 BLEU score

- Utilized Transfer Learning with ResNet-18, CNN models to classify CIFAR-10 images with 88% accuracy

- Built multi-layer & bidirectional character-level RNNs to classify names nationality with 86% accuracy

- Predicted lending club loan default with 0.89 precision & recall both: lead to a loan selection strategy that invests in 1.76% of available loans with 15% annualized return

*Full-time summer graduate assistant in UIC on AI assignment for client ClarisHealth **Under revision

Contact this candidate