CS Graduate | ML Research | Seeking full-time opportunities | Arizona State University
Experience
CCC Information Services, Chicago, IL : R&D/Data Science Intern May 2019 – Aug 2019
• Worked with the R&D team on the Mobile Crash Detection project. The data used for this project was a combination of Telematics data and mobile sensor data. • Developed a machine learning model to detect phone drops. (Python, Scikit-learn). Utilized an ensemble model to make weighted predictions. • Performed feature engineering to reduce the false positive rate. This task required an extensive use of feature space analysis. • The machine learning model improved the performance of the overall model by significantly reducing false positives by 30%
Curl Analytics, Bangalore, KN, India: Data Science Intern Jan 2018 – Jun 2018
• Developed Vehicle Counting System using YOLO Deep Learning model to enable traffic authorities to monitor the traffic. Techniques like Transfer Learning were used to improve performance. • Utilized Deep Learning models like Single Shot Detector and Resnet to compare the accuracy and speed of various models. YOLO model was found to have significantly higher accuracy at mAP = 57.9%. • Using OpenCV, implemented a tracking module with techniques like LK optical flow and Canny Edge detector.
Projects
Published a NLP research paper - Exploiting Emojis for Sarcasm Detection (https://www.researchgate.net/publication/332082181_Exploiting_Emojis_for_Sarcasm_Detection)
• Text and Emoji data from Facebook and Twitter comments were scraped and categorized. (Python) • Word Embeddings and Emoji Embeddings were generated using Word2vec and Emoji2vec. (NLTK) • This data was used to train and test the Bi-Directional LSTM Attention model to perform classification. (Keras)