I'm a perfectionist. There, I said it. As a life-long musician, perfectionism is an intrinsic part of my personality now - something that has overlapped into other facades of my life as well. Its a quality that goes hand-in-hand with patience, relentless obsessiveness, and sleepless nights; all in all, a perfect combination for a grad student!
I love data science - I have dabbled with data science projects in PySpark/Spark, Python, MATLAB, and R. My experience interning at Mad Street Den has given me a lot of exposure to writing MapReduce jobs in PySpark/Spark.
3800 SW 34th Street, Apt EE 309
Gainesville, FL 32608 US
+1 (949) 505-9907
Master of Science in Computer Science • graduating May 2017
Grade Point Average (GPA): 3.67
Courses taken: Analysis of Algorithms, Distributed Operating Systems, Data Mining, Human Computer Interaction, Pattern Recognition, Advanced Datastructures, Statistics, Bioinformatics, Computer Networks.
Organizations part of: UF Data Science and Informatics (UF-DSI), Association for Computing Machinery (ACM), Association of Computer Engineers (ACE)
Bachelor of Engineering in Electronics and Communication Engineering • Aug 2011 to June 2015
Data Science Intern • April 2016 to September 2016
• Worked primarily on PySpark/Spark, and Python. Designed and implemented an end-to-end NLP project using PySpark, by first building a customized tagger for product descriptions using CRF and feeding this into separate word2vec models, and finally classifying the product based on style and occasion.
• Have also contributed to a map-reduced version of the k-medoids algorithm in PySpark, performed benchmarking tests on PySpark and implemented a unit test framework in PySpark.
Independent • July 2014 to November 2014
Creation of Complex Percussion Patterns in Indian Classical Music Using Sparse Representation Learner
Proposed a method to generate complex melodic beat patterns using feature extraction, by mapping melodic patterns to vectors which are then given as input to the Sparse Representation Learner. (paper accepted at IEEE Spices 2015).
Python, PySpark, Spark, Scala, Java, C, C++, R, MATLAB, C#
InVision, Adobe Creative Suite, Git
STL Libraries, OpenCV, cvx-opt, scikit-learn (Python), pandas (Python), akka toolkit (Scala), spray toolkit (Scala), gensim (Python), NLTK (Python), textblob (Python)
Python, PySpark, R, pandas• Apr 2016
• Implemented a project that uses the Yelp Dataset to predict trends and future business attention, and identify potentially lucrative business locations for different business categories.
• Designed a mobile/web application to onboard users to new cultures and mitigate culture shock through gamification.
Java • Mar 2016
• Implemented an Event Counter using a Red Black Tree in Java, which reads redirected input from a text file, executes different commands on the data, and prints the results.
• The program was tested successfully with an input text file containing 100 million records.
• Designed a prototype of a mobile news app called 'HappyFeed', based on user research.
Scala • Dec 2015
• Designed a client-server simulation model of the Facebook REST API supporting over 10000 users using spray-can and akka actor model and simulated features like profile, friend list, photos and albums.
• Implemented end-to-end security using a combination of AES and RSA public key cryptography techniques.
R, Python • Dec 2015
• Created a tool as the lead team member to predict genre of movies from a dataset extracted from IMDb containing 17000+ records. Performed text mining on the dataset to build new features.
• Developed routines to perform classification, find association rules and cluster similar movies together.
Scala • Oct 2015
• Implemented gossip, push-sum algorithms and chord protocol with an object access service.
• Calculated and analyzed the convergence rate in different topologies for gossip and push-sum algorithms.
Scala • Sep 2015
• Implemented Bit Coin Miner using akka actor model in Scala
MATLAB • Mar 2015
• Developed an application to perform OCR for three Indian languages (Tamil, Telugu, and Hindi), using NMF as the dimensionality reduction tool and Sparse Representation Learner as the classifier.
• Created customized feature extraction techniques to map the character images from matrices to vectors.
Research paper accepted at IEEE Spices 2015 • Nov 2014
• Proposed a method to generate complex melodic beat patterns using feature extraction, by mapping melodic patterns to vectors which are then given as input to the Sparse Representation Learner.
VBA • July 2014
• Developed a tool to perform Behaviour Profile Assessment for a large number of subjects and to analyze the data collected in detail, using the DISC model of four factors, subsequently printing the results as a report.
Sudarshan Babu and Vikaasa Ramdas, "Ensembling Sparse Representation Classifiers through Layers of Support Vector Machines", to be published in the proceedings of the 15th IEEE International Conference of Machine Learning and Applications (ICMLA), Anaheim, USA 2016.
• Hosted a PySpark Workshop in Vellore Institute of Technology (VIT), on August 13th 2016.
• Built a project to classify speeches by presidential candidates Barack Obama and Mitt Romney using PySpark, to demonstrate to the students the different aspects of building an end-to-end NLP project. These included teaching the students data cleaning and munging routines as a series of map and reduce operations, extracting n-grams, building word vectors, and finally, implementing the pipeline for the machine learning algorithm.
• Explained to the students PySpark concepts related to broadcast variables, accumulators, and writing UDFs for Spark Dataframes and map functions for RDDs.
• Also gave a brief overview of setting up AWS EMR cluster for running PySpark jobs and defining the different user configurations for setting up the cluster.
'HappyFeed' is a unique interactive social news application that is personally customized to provide you with positive news that interests you.The design of this app was based on user research. Click on 'Details' to view the InVision prototype.Mobile App Design, Prototype, User Research
This project uses the Yelp Dataset to predict trends and future business attention, and identify potentially lucrative business locations for different business categories. First, heirarchial clustering is performed to cluster geographic clusters of businesses. Then, the Random Forests regression model is used for predicting future business attention, and association rule mining is done to dentify complementary business categories. Finally, clusters with a large number of complementary businesses and low number of competing businesses are identified, which indicate potential business opportunities.Python, PySpark, R, pandas
I'm an Indian Classical instrumentalist playing the electric mandolin for over 16 years, and am a disciple of the renowned Mandolin U. Shrinivas and his brother Mandolin U. Rajesh.
Do check out my YouTube channel and my website, www.mandolinramdas.com