Career Profile
I am currently pursuing Masters in Computer Science & Engineering at University of California, Santa Cruz. I have a strong Algorithms and Data Structures foundation along with Three years of Software Engineering experience. I am highly skilled in Python, SQL, Java and Unix having worked on it for the past Five years now. My research interests are in Data Science and Natural Language Processing. My non-academic interests include competitive coding, hiking, backpacking and photography. I have strong interpersonal and leadership skills. I am also a theater enthusiast.
Experiences
- Led chatbot development projects for two major banks in the Philippines, currently used by over 30,000 users.
- Built dialog flows and integration adapters with third-party platforms like Facebook Messenger, Slack, Microsoft Teams.
- Implemented context management, intent and entity recognizers using RASA NLU & Microsoft LUIS and created telemetry analytics dashboards.
- Incorporated statistical and machine learning models in the Unscrambl Drive platform for customer segmentation, user profiling, variance analysis, purchase propensity calculation for the biggest Telecom in the Philippines.
- Constructed complete data pipelines for data preprocessing, aggregating, analyzing and generating visualizations for various clients.
- Classroom mentor and project reviewer for “Udacity Deep Learning Foundations Nanodegree”.
- Mentored 100+ international students for the course including Q&A sessions with the students regarding various projects and course material.
- Reviewed and graded 100+ assignment projects related to the course.
- Worked on a major release for enhancing an existing commercial online banking software using Java-related technologies in the forward development team.
- Added a feature of dynamic reloading of modified spring components in the application. This eliminated the need for a complete application server restart after any modifications were made to the spring code. The change deployment time reduced from about an hour to 3-4 minutes.
- Developed and maintained an automated REST API testing framework for a banking application.
Projects
Some projects which I have worked on
Face generator
- Trained a generative adversarial network using tensorflow to generate new faces from celebA database. Languages/Frameworks used: Python, Tensorflow, Jupyter Notebook
Content-based audio auto-tagger using deep learning
- Tackled the multi-label classification problem of accurate prediction of tages for audios. The experiment was performed on MagnaTagATune (MTT) dataset using Convolutional Neural Network (CNN) and Convolutional Recurrent Neural Network (CRNN). An AUC-ROC score of 0.886 was achieved. 40 fixed tags in emotions, instruments, genre could then predicted for new tracks.
Neural network from scratch
- Developed a complete neural network from scratch using numpy to understand the fundamentals .Implemented the Forward and Backword Propogation passes, wrote my own implementation of sigmoid which was used as an activation function and trained the network on bike dataset to find the optimum number of bikes required each day. Tweaked the hyper-parameters to understand the variations and effects on the accuracy of the network.
Meta-App | A framework for incognito mode for smartphone applications
- An Intent Publishing solution which can be used by the user to request offers from dealers anonymously.
Centralized student database
- Setup a Hadoop cluster to act as a centralized location to store student information (attendance, marks and teachers' recommendations and comments about the student) in JSON objects. It could be useful when student is switching classes or schools to convey a better student profile over traditional quantitative attributes.
Publications
I plan to do more focused research and publish more papers during my Masters
Content-based auto-tagging of audios using deep learning
International Conference on Big Data, IoT and Data Science (BID), 2017
Meta-App A Pull-Based Approach
National Conference on Advancements in Computer and Information Technology, 2016