I am currently working on building multi-lingual models for moderating ads for Amazon Sponsored Products. Previously, I have worked on Reinforcement Learning (RL) and its application over a variety of problems, such as supply chain, optimization, cost to serve, dialogue generation etc. I also looked at making RL algorithms more robust and sample efficient using predictive knowledge inculcation. I have also worked in the Intersection of Natural Language and Deep Learning specifically for sentiment analysis. As a side projects I have dabbled into analyzing social media text from a statistical and behavioral perspective.
My Research Interest lies in
1. Intersection of Natural Language Processing with Deep Learning and Reinforcement Learning, including modelling behavior patterns
2. Sample Efficient Reinforcement Learning
3. Artificial General Intelligence
I have had the opportunity to work across domains in machine learning which has helped me to observe ML/DL space from various vantage points. As one of the pioneers in computer science, Alan Ray said “Change in perspective is worth 80 IQ points”, perspective does help in solving problems.
https://hardikmeisheri.github.io/