Machine Learning and Behavioral Research Intern Seeks Opportunity
1. TARANVEER SINGH
taranves@andrew.cmu.edu | +1 (412) 954-7029 | www.linkedin.com/in/taranveer
OBJECTIVE
An internship or research opportunity that exposes me to behavioral research methods and machine learning models on real word datasets.
EDUCATION
Carnegie Mellon University Pittsburgh, PA
Masters in Computational Data Science December 2017
Major in Human-Centered Data Science
GPA: 3.72
Vellore Institute of Technology Vellore, India
Bachelor of Technology in Computer Science May 2016
GPA: 3.61
Relevant Courses: Introduction to Computer Systems (15-213), Design of Human-Centered Software, Machine Learning (10-601),
Applied Econometrics, Interactive Data Science*, Machine Learning for Text Mining*, Measuring casual effects in Online Platforms*
*Ongoing
SKILLS
Programming Languages: R, Python; C/C++, SQL.
Python Libraries: numpy, pandas, sckit-learn, nltk
EXPERIENCE
Centre for Artificial Intelligence & Robotics Lab (CAIR) Bangalore, India
Research Assistant Jan 2016- May 2016
Implemented classification based system that could differentiate between subjective and objective sentences in a news article.
- Designed and implemented data pipeline to generate the training data using Guardian API and Opinion miner.
- Extracted Ngrams to find probability of sentence being a subjective and objective sentence.
- Used word-sense disambiguation and general inquirer dataset to find context of a word leading to subjectivity.
- Used xgboost classifier on the degree of subjectivity and word context to classify sentences into subjective and objective with a
F1-score of 60%.
Fitrrati – E-commerce Based Startup Delhi, India
Data Science Intern Mar 2015 - Dec 2015
Implemented Data Pipeline for Size Recommendation System Module
- Implemented Outlier Detection on size measurement charts by a mixed model of IQR & Multivariate Probability Distribution to
flag anomalous charts.
- Examined garment clusters to visualize disparity in size chart measurements across different brands, product and fit type.
- Performed Associative Rule mining to figure out most frequently entered User size measurement input pattern.
- Used Multivariate probability density function to mark anomalous user measurements (inputs) for verification.
- Used PCA to find latent variables for clustering of users based on user body measurements.
BondForever – Dating Startup Bangalore, India
Data Science Intern Sep 2014 - Jan 2015
Implemented Match-making algorithm to recommend matches based on user preference.
- Established a compatibility score to quantify compatibility of user profiles.
- Designed and implemented Hungarian algorithm based Recommendation System to find best matches.
- Automated user approval algorithm to send matches only when sufficient pool of quality of matches are available for the user
based on his profile.
PROJECTS
Question Selection using Item Response theory (IRT) and Relative Ability Estimate (Github)
The project was aimed to select questions from question bank for exams that give good estimate of relative ranking between students and
by estimating question discrimination i.e. how well a question is able to separate low scoring from high scoring students.
- Estimated Question Discrimination values using EM based on 2 PL model of IRT and Relative Ability Estimate via Collaborative
Filtering.
- Identified ambiguous questions from the question bank with negative discrimination.
- Maximized the spread of questions over all possible latent ability of students.
PUBLICATIONS
Raghuveer V. R., B. K. Tripathy, Taranveer Singh, and Saarthak Khanna "Reinforcement learning approach towards effective
content recommendation in MOOC environments." In MOOC, Innovation and Technology in Education (MITE), 2014 IEEE
International Conference on, pp. 285-289. IEEE, 2014