I recently graduated from University of Southern California, Viterbi School of Engineering with a Masters in Computer Science. I have project backed expertise in Front-End Development, Android Studio, Machine Learning and Data Mining. I'm passionate about Web Designing and have taken it up as a personal hobby. With a broad skill set in the Computer Science field, I'm looking for opportunities in the same.
-
Experience
I worked as Machine Learning Intern for Directed Research at University of Southern California. I worked on CarmaCam, a project developed to identify vehicles violating traffic rules on road using the video captured by the person filing complaint of the respective vehicle on the app. My role is to determine vehicles that are speeding, analyzing the requirements and proposing the best model. I drafted a model using OpenCV to detect over speeding vehicles in videos and identified vehicles labels using YOLO object detection algorithm and programmed speed calculator using detection box displacement and frame timing. Also, working on the project efficiently and meeting the deadlines thereby improved my time management skills. Working in a team and collaborating the work with the other teams, gave me an experience of developing a project in modules.
-
Projects
Responsive Weather Forecast Webapp V.1.0 using (PHP, JavaScript, HTML, CSS) | V.2.0 (Angular 7, NodeJS, HTML, CSS, Bootstrap)
I Designed and deployed a dashboard on Azure to forecast weather for locations in US using DarkSky API and the retrieved results comprised of multiple weather parameters with graphs for the following week like temperature, ozone, humidity, wind speed, precipitation, etc. I designed a version 2 of the same application in which I routed Google and DarkSky API calls through NodeJS server setup on Azure. I also made the design responsive using Bootstrap and added favorites tab feature for multiple locations.
Solr Web Search Engine using (Solr, Python, PHP, HTML, CSS )
I built a search engine to display results of NYTimes.com inclusive of autocomplete and spell check features. I also incorporated Lucene and PageRank algorithm using inverted index of downloaded HTML files of NYTimes, created by Solr and TIKA parser
Food Truck Android App using (Android Studio, Java )
I crafted an application to find open street food trucks in San Francisco along with map. I initiated a description box with address, operating hours and items available for particular food truck, on click of respective location marker in the map view
Yelp Recommendation System using (PySpark, Python)
i weighted features of Yelp dataset against each other to maximize accuracy. I selected user, average stars of business, review count, latitude and longitude as the best attributes to predict star ratings. I achieved a RMSE of 1.07 on predicted user ratings on comparison with a trained user-based model. I enhanced the model with a hybrid combination of XGBoost and SVD algorithms and accomplished a higher accuracy and lower time taken with a RMSE at 0.981