Harshal Chaudhari Personal Information Date of Birth Contact Website May 14, 1991 harshal@bu.edu +1-857-294-8568 http://cs-people.bu.edu/harshal Research Interests Statistical data analysis in E-commerce, Bayesian Learning Education 2015–current Ph.D. in Computer Science, Boston University Research Adviser : Prof. John Byers GPA : 4.0/4 2013–2015 2009–2013 M.S. in Computer Science, Boston University B.E.(Hons.) in Computer Science, Birla Institute of Technology and Science, Pilani Research Experience Fall 2015 Impact of Zero-price Promotions on Sales: A case study of Amazon Appstore, Ongoing Supervisor : Prof. John Byers Amazon, after making a late entry in the app economy, has captured considerable marketshare by providing attractive deep discounted promotions like ‘Free App of the Day’. Even though such promotions offer highly sought after visibility to the participating developers, they only benefit if the incremental post promotion sales offset the cost of ‘free giveaways’ on the day of promotion. In this work, I perform a rigorous statistical analysis to show that exposure received during such promotions does not always lead to incremental future sales. Moreover, the utility of the promotions strongly depend on the app characteristics themselves. Fall 2014 Identity discovery challenge Instructor : Prof. Peter Chin In a course project on Compressed Sensing, we aimed at using compressed sensing techniques to identify a potential disease carrier from a sparse healthcare dataset, using his incomplete record. We used Bayesian mixture models and Expectation-Maximization to identify overlapping communities. Identifying the community to which the partial record belongs to reduces the search space by an order of magnitude. Within this community, we used shortest paths between attributes of the incomplete record to identify a set of potential disease carriers. Spring 2014 De-anonymizing users across social networks Supervisors : Prof. John Byers, Prof. Evimaria Terzi We evaluated the efficacy of privacy measures on social networks; we conjecture that it is possible to map users across different networks using the similarity of the induced subgraphs of users irrespective of their privacy settings. We used de-anonymizing coverage of a social network based on its look-ahead to compute a lower bound on the number of completely observable users required to de-anonymize the entire network and validated it on synthetic power-law graphs. To verify results on real datasets, I extensively scraped Quora and Instagram, by reverse engineering XHR requests to scrape dynmaically loaded AJAX content. I also gave a talk on techniques of data collection and organization for computational journalism at the “Story Telling with Data” conference at Boston University. Spring 2012 Cluster Labeling Using Term Ranking Supervisor : Prof. Rajendra Roul Modified Google’s PageRank algorithm to incorporate undirected edges and edge weights in a relational graph and segregate keywords of a cluster into categories of discriminative, ambiguous and common terms. The high ranking discriminative terms are used to extract generic cluster labels from the external sources such as pre-categorized knowledge of Wikipedia. Work Experience Summer 2015 Personalized Marketing User Simulator, Globys Inc., Seattle, USA Developed a user simulator for personalized marketing platform, Amplero, developed by Globys Inc. The characterstics of the simulated users are derived from real world telecom usage data in a probabilistic manner. It removes operational lag associated with marketing, facilitates A/B testing for various predictive models devised by Amplero. Modular design of the user simulator further enables fine tuned differential analysis of the strategies. Spring 2013 Clone Prioritization Algorithm, Siemens Corporate Research, Bangalore, India Developed a strategy for identification of code clones and optimal resource allocation for clone refactoring. Modeled this as a multi-constrained, multi-objective Knapsack problem and investigated various heuristics in multi-criterion branch and bound algorithms in addition to evolutionary algorithms based on Pareto optimality. We developed an Eclipse plugin for the assessment of code duplication characteristics. In my leisure, I also implemented a customized version of GIT on SCR servers. Summer 2011 Database Management System for Production charts, KPCL, India Developed a PHP API and a Web application for migration from legacy data storage to a centralized database system. The data extracted from the gearbox production design charts is presented in an organised sequential manner to the managers of the production process. This software is being used at Kirloskar Pneumatic Company Ltd., India (KPCL) and has increased the efficiency by reducing production time by over 20%. Teaching Experience Fall 2014 Summer 2014 Fall 2013 Spring 2014 CS131 : Combinatoric Structures, Instructor : Prof. John Byers The course provides a strong background in mathematics to computer science majors. My responsibilities include designing questions for lab sessions, leading discussion hours in addition to grading exams. CS111 : Introduction to Java Programming, Instructor : Prof. David Sullivan I am responsible for lab sessions, discussion hours, grading of over 200 students along with review lectures. At the end of the year, I was rated 4.7/5 on an average for my teaching capabilities by over 400 students. Key Academic Projects Spring 2014 Modelling Königsberg Problem in Alloy Analyzer We used Alloy specification language to model the famous Königsberg problem. Our model verified the predicate that no Eulerian path exists in an undirected graph representing the seven Königsberg bridges. We further generalized the model to show that an Eulerian path exists if and only if every vertex of the graph has an even degree. Fall 2013 MemOS - Operating System Programmed a barebone OS that boots via GRUB and probes system BIOS. It features a preemptive round-robin thread scheduler functionality, implementation of a custom filesystem based on EXT2 and its corresponding file operations. Implementation in C and assembly languages. Fall 2013 Multi-threaded Streaming Video Server Implemented a multithreaded server that streams sequences of Star Wars video frames, via sockets, to client threads. It also supports text based client requests. Advanced part of this project involved adding monitor abstraction and synchronisation routines in the kernel space. Fall 2013 Author Disambiguation Implemented duplicate author detection over a Microsoft Academic Search database of 2 million authors. Identification criterion uses metadata such as publication topics, affiliations and co-authors along with fuzzy name matching. Implemented in a mutithreaded program that iteratively maximizes precision and recall. Autumn 2012 Android Lens Developed an Android application which recreates the universal search functionality of Ubuntu Unity dock. It combines the search functionality of major websites like Wikipedia, YouTube, Amazon, eBay, etc. as well as local phone media. User has the ability to dynamically add more search Lenses via Google Play. Spring 2012 Convex Polygon Parser As a part of Compilers course, a lex based convex polygon parser was developed. The parser takes bitmap image as its input and identifies the number of sides of a polygon. Hobby Robotics Projects Summer 2010 Micromouse, a maze solving robot Developed an autonomous maze solving robot which uses the A∗ algorithm for path planning. Novel techniques like chopper control of stepper motors, modeling inverse kinematics are used to achieve better performance. Programmed an algorithm simulator using GTK and a RF debug interface for development. Spring 2009 Line Grid Solver Developed an autonomous line grid solving robot programmed to find the least expensive route to the goal in an unknown grid using Dijkstra’s algorithm. It uses control systems feedback techniques like PID with inverse kinematics for proper positioning of the robot. Awards and Scholarships Academic 2007–2012 2005–2007 2005–2007 National Talent Search Scholarship by the Government of India. Maharashtra Talent Search Scholarship by the Government of Maharashtra state. Bombay Talent Search Scholarship. Miscellaneous 2004 1st place in All India Camel Color Contest. Over 100 awards at state and zonal levels in painting, chess and quizzes. Extra-curricular activities 2011–2012 Chief Designer, Department of Arts and Decoration, BITS Pilani I was responsible for various art related activities of the institute. I designed logos and branding material for institute cultural, technical and sports events. 2011 Co-ordinator, Kala, Fine arts club of BITS Pilani I played a pivotal role in the foundation of the club and organised art events in the institute. I also conducted a series of workshops on Digital Art and portraiture. 2009–2011 Chief Artist, Department of Creative Works, BITS Pilani I was responsible for designing and development of websites of institute festivals, conferences and other departments. I conducted workshop on Website Designing and Development. Skills Programming Softwares Python, Java, MATLAB, R, C/C++, Javascript, Php, MySQL. OpenCV, Xilinx, Verilog VHDL, Omnet++, Pspice, Eagle CAD, Autocad. Recommendations Prof. John Byers Professor, Department of Computer Science, Boston University, Boston. Ph. No. +1-617-353-8925 byers @ cs.bu.edu Prof. Evimaria Terzi Associate Professor, Department of Computer Science, Boston University, Boston. Ph. No. +1-617-358-5851 evimaria @ cs.bu.edu Dr. Oliver Downs Chief Technology Officer and Chief Scientist, Globys Inc., Seattle. Ph. No. +1-206-352-3055 odowns @ globys.com