KESHIN JANI
•Tempe, AZ • +1-480-***-**** • adb1at@r.postjobfree.com •linkedin.com/in/keshinjani/
EDUCATION
Master of Science in Computer Science, Arizona State University Graduating in May 2021 Courses: Distributed Databases Systems, Multimedia and Web Databases, Fundamentals of Algorithms. GPA: 3.5/4.0 BTech in Computer Engineering, Mumbai University Graduated in July 2018 Courses: Data Structures, Algorithms, Database Systems, Operating Systems, Computer Networks. GPA: 8.2/10.0 TECHNICAL SKILLS
Programming: Java, Scala, Python, Shell Scripting
Frameworks and Tools: Apache {Impala, Spark, Kafka}, OpenCV, TensorFlow, Django Databases: PostgreSQL, Apache (Kudu, Hive, Cassandra}, MySQL WORK EXPERIENCE
Graduate Analyst Barclays, Pune, India July 2018-July 2019
• Speeded up fraud analytics by getting real time data amounting to 12 million records per day per entity
• Aided marketing team by getting transactional data on real time basis and enhancing the real time data analysis
• Developed a framework consisting of housekeeping, monitoring, audit and reconciliation of data received through the different Spark applications.
• Collaborated with other teams to help them utilize Kafka services into their applications Web developer ISTE Council, Mumbai, India June 2016-June 2017
• Developed and built a website to showcase events and workshops conducted by the council
• Created a site for State Level competition including technical paper presentation and working model project exhibition
• Provided technical expertise for the workshops and hackathons. ACADEMIC PROJECTS
Image search database Fall 2019
• Implemented Personalised PageRank, Decision trees and SVM from scratch to classify similar images in a database
• Implemented Locality Sensitive Hashing in high dimensional space, reducing the similar images retrieval time by 35%
• Implemented feature reduction techniques like SVD and PCA to analyse underlying latent features in image data for efficient storage and classification
Meal Identification Fall 2019
• Extracted multiple features from the carbohydrate data of patients and trained 4 different classifiers – SVM, Logistic regression, Random Forest and perceptron with maximum accuracy of 83% of accuracy on test data
• Clustered the extracted features into multiple clusters using K- means Troll Detection System Fall 2017
• Transform the comments into binarized features using TF-IDF
• Developed model for every toxic class: toxic, severe toxic, obscene, threat, insult, identity hate using Naïve Bayes and Logistic Regression
Presented a paper on the same in an IEEE conference - 3rd International Conference for convergence in technology Inventory System Fall 2016
• Built a system to record and visualize different variety of stocks present in warehouse
• Utilized PHP and MySQL to implement functions to record all the incoming and outgoing stocks