Hello, I am Raja, a computer science master graduate specializing in neural machine translation. I am intereseted in NLP and Deep Learning in general. My research focus is on improving maching translation for morphologically rich languages.
Improving Neural Machine Translation for morphologically rich languages (MS Thesis)
Proposed and implemented an NMT architecture in PyTorch that can improve the translation quality of morphologically rich languages like German, Turkish, Tamil etc by incorporating subword information.
Successfully improved the translation performance on German->English and Turkish->English by 2-3 BLEU points against the benchmark transformer model
Unsupervised morphology induction using word embeddings
Implementation of the Unsupervised morphology induction paper for CMPT882: Neruo Machine Translation course at Simon Fraser University
Implemented morphological induction technique from the paper and used it to study the performance used it to study the performance of different word embedding models on their ability to capture the semantics of the words.
India (2012 - 2015)
After finish my undergrad Engineering at MIT, Anna University, Chennai in Computer Science, I started my professional career at 2012 as a Network Engineer at AGC Networks. Worked on implementing various end to end high availability networks consisting of APs, Switches, Routers, Controller and Firewalls. Promoted as Lead-Projects. Handled a project from Project plan to Project Sign off.
I joined Athenahealth Technologies, a cloud based service provider for health care at 2013, where i worked on web applications in Perl, JS.
Then, went into self-discovering mode for year and tried my hands on travel startup in 2014. We built a personal travel assistant that was the one stop destination for all you travel needs starting from travel planning, booking to the actual travel. Handled complete development process including architecture design, frontend and django backend. Closed the company for lack of funding and started my Master at UNBC, Canada in Computer Science.
Canada (2015 - 2018)
While persuing my master at UNBC, I worked on a strategy based learning startup called Road2Math. I designed and development web applications using AngularJS, socketIO with NodeJS backend.
Parallely, I worked as and Android Developer in a Publishing company called Angler’s Atlas. I also developed a research project (Python) which extracts locations specified in texts from social media platforms like youtube description and geo tags them using NLP and Machine learning algorithms. It was essentially an NER problem.
In 2017, I joined Timber Tracks Inc. where I currently work full-time as Full Stack Developer. We develope a management / analytical tool to improve productivity of logging companies. I develop web app (AngularJS), mobile app (ionic) and backend (Django)