Shaun D'Souza has around two decades of experience in AI, ML, Software Engineering, R&D, Business. He has a varied experience in Computer Science and Engineering. Shaun completed his BS from Cornell University with a Double Major in Computer Science, Electrical and Computer Engineering, Cum Laude Honors in 2003. He completed his MSE in Electrical Engineering from the University of Michigan, Ann Arbor, Electrical Engineering and Computer Science Department in 2005. Shaun has worked in Software/Business researching machine learning, generative AI, compilers, algorithms and systems. He has published several papers and been granted patents.
Shaun has developed ML algorithms for supervised learning, unsupervised learning and reinforcement learning. He has designed and implemented a compiler for an object-oriented language similar to Java. Shaun is exploring a broader sandbox ecosystems enabling for AI. He is investigating ML algorithms on unstructured content form on the web. He is evaluating the intersection of Software and Business in the context of AI, Globalization, CSR and the Last mile with an emphasis on ML algorithms in the broader web.
As part of his Graduate research, Shaun worked in the Advanced Computer Architecture Laboratory at the University of Michigan, Ann Arbor. They developed a full-system architectural simulator called M5 in C++ and Python. Their research on PicoServer was published in the Architectural Support for Programming Languages and Operating Systems (ASPLOS) conference. Additionally, he worked on developing pronunciation scores for phoneme, word and phrase using Hidden Markov Models (HMMs) and statistical language models (Ngram) based on textual data from the Gutenberg project for speech recognition on CMU Sphinx. Shaun has developed ML Algorithms in Matlab for supervised learning (neural networks, naive bayes, decision trees, support vector machines), unsupervised learning (k-means clustering, dimensionality reduction) and reinforcement learning.
Cornell's Computer Science curriculum is primarily taught in the Java programming language. As part of their CS project course, they developed a compiler for an object-oriented language similar to Java called Irish Coffee. The Compiler was developed in Java with more than 15K LOC to support lexical analysis, parsing, type checking, optimizations and code generation. Shaun has completed coursework in Machine Learning, Advanced Compilers, Algorithms, Graphics and Operating Systems. He is a recipient of the Gates Millennium Scholarship from the Bill & Melinda Gates Foundation.
Shaun has worked with some of the top companies in the Software industry including, Accenture, Wipro, Tata Group, Intel and IBM. On graduation, he interned with IBM in the Sony Toshiba IBM Design Center. At Intel, he worked in the Ultra Mobility Group. At Wipro, he worked in the CTO Office on HOLMES. Shaun has published papers and been granted a patent on Information Extraction and Syntactic Parsing. The Granted Patent in AI, ML (NLP) is on a System and method for extracting information from unstructured text. He received Best Pragati at Wipro for his poster presentation on Eclipse CDT code analysis and unit testing. The paper was PeerJ's Top 5 most viewed article in Programming Languages, 2018. His papers on Evolving system bottlenecks in the as a service cloud, Cognitive Architecture for a Connected World and Parser extraction of triples in unstructured text are published in arXiv.
Shaun has published a paper on Holistic generational offsets: fostering a primitive online abstraction for human vs. machine cognition in arXiv. He has filed patents on Sentence phrase generation, Textual entailment and an Artificial intelligence and machine learning based conversational agent. His paper on Defining a Sandbox for Responsible AI was in SSRN's Top 100 in CompSciRN: Artificial Intelligence, 2018. His paper on LSTM neural network for textual ngrams was SSRN's Top Ten download list for CompSciRN: Artificial Intelligence, Dec, 2018. He has presented in Knowledge Sharing Sessions, Expert Talks and Brown Bag sessions on AI and ML.