Apr 2021 – Feb 2026 · 4 yrs 10 mos

Feb 2021 – Feb 2026 · 5 yrs · Bangalore Urban, Karnataka, India

Jun 2018 – Feb 2020 · 1 yr 8 mos · Vishakhapatnam Area, India

Nov 2016 – Feb 2021 · 4 yrs 3 mos · Bangalore

Dec 2014 – Jul 2016 · 1 yr 7 mos · Greater Marseille Metropolitan Area

• I am part of Sensei-conversation project where I work on areas related to conversation analysis. My work focuses:
• 1. On algorithm and technology development for interactive and visual summarization of large scale conversational textual data
• 2. Developing use cases for making large scale conversations more understandable
• 3. Using machine learning based methods to monitor the quality of call center conversation
• 4. Using Deep learning based methods for detecting semantic change over large corpora
• 5. Connecting with other sensei- partners and enabling technology transfer for prototype development

Feb 2013 – Oct 2014 · 1 yr 8 mos · Bengaluru, Karnataka, India

• SAIT Labs functions
• 1. Developing mobile based state-of-art sentiment analysis engine
• 2. Methodologies for automatic evaluation of crowd sourced data
• 3. Andromorphic mobile technology use case development
• 3. Publishing and patenting
• Social Live Imaging Projects
• 1. Creating NLP related technologies for Interest mining
• 2. Leading team to build interest mining use cases
• 3. Ideating disruptive applications for Samsung galaxy and note series of mobile phones

Sep 2012 – Dec 2012 · 3 mos · Bangalore

• Research project to find the relation between dependency parsing and machine translation for Asian languages.

Jul 2008 – Dec 2012 · 4 yrs 5 mos · Mumbai Metropolitan Region

• I was part of the Sentiment Analysis (SA) Team of IITB. My study focused on resource constraints related to SA for Indian languages. I used different machine learning algorithms to boost the available data and then perform SA. One of the take away from my thesis is use of abstraction at feature level to improve SA performance. This abstraction is carried by projecting the the lexical feature space into semantic feature space. Using this technique, SA can be performed on most of the official Indian languages with minimal data.
• Another problem I address in my thesis is domain adaptation for SA. An Ensemble of strong classifiers with intermediary training based on noisy data seems to solve the domain based constraints related to SA.
• Apart from these, I worked on real time sentiment detection from Twitter. As a part of this, a streaming based SA engine was created. The objective of this endeavor was to use the information provided by this engine to do predictive analysis. One interesting task related to this is finding the financial fluctuation based on sentiment.
• As a part of coursework related to PhD, I worked on many interesting Artificial Intelligence (AI)and Machine Learning related problems.

Jun 2005 – Jun 2008 · 3 yrs