My Portfolio

Converged an optimal set of configurations for a given IoT deployment to reduce energy consumption while satisfying required threshold accuracy and maintaining the quality of service.

Implemented Linear and Multiple regression, Polynomial curve fitting and regression for Univariate and Bivariate data sets. Also implemented classifiers like Bayes, Perceptron based, Logistic regression- based and SVM for Linearly separable, nonlinearly separable and Overlapping data sets.

Developed website using HTML, CSS, PHP (frontend) & MySQL (backend) with relational databases (airlines, timing, destinations, ticket cost) & hosted it locally. Enabled users to enter the source and destination boarding points and book flights using the same tools.

Trained a Named Entity Recognition (NER) model for Natural Language Processing (NLP) of clinical health records. Collected & cleaned as per application requirements using Python. Implemented NER models from SpaCy & SciSpaCy libraries, compared their accuracies by varying hyper-parameters. Observed SciSpaCy en_ner_bc5cdr_md model with pre-training and achieved more accuracy with an F1 score of 0.846 after 100 iterations.

Developed application to enable efficient communication between parents & teachers with features like chat support, performance analysis, grading, etc. Built frontend using Google Android Studio (Java, XML) and Git Labs for version control; used Database Administrator to store student data (names, marks, blood group, grade, etc.). Implemented Restful API using Retrofit & Volley libraries that use HTTP requests to GET, PUT, POST and DELETE data from Server.

Designed a novel intelligence scale using Wavelet Packet Transform for feature extraction & Hierarchical Extreme Learning Machine for classification. Tested the scale on students & analyzed EEG signals from the frontal cortex of the brain using an EEG cap and amplifier while answering a quiz . Collected & classified students as low , high intelligence, etc. based on data analysis in terms of ~256 data points. Achieved 80% training accuracy and 73.33% testing accuracy.

Over the course of this project, I studied numerous search algorithms, and applied them to a practical use. Implemented Depth First Search, Breadth First Search, Uniform Cost Search, A* search in order to reach and eat the precoded fruit. By comparing all 4 methods, we found that A* had a better performance as it computed the path cost with a heuristic.

In this project, our task is to build a web search engine Building a crawler to scrape data from a specific domain Indexing and Ranking the scraped data using Lucene Indexing and Ranking the collected data using Hadoop Building a web interface to demonstrate the functionality of oursearch engine