Away from the desk.
APPLIED SCIENCE · MACHINE LEARNING
Hi, I’m Hardik.
Trying to build better models.
Learning a little more with each iteration.
I’m a Senior Applied Scientist at Microsoft AI, where I work on foundation models for advertising. My research spans reinforcement learning and NLP, with applications in supply chains, games, and sentiment analysis.
Before Microsoft, I worked at Amazon Advertising and TCS Research. Here you’ll find my papers and notes on what I’ve been learning.
Selected research
Browse by topicPick a topic to browse the papers. Some work belongs to more than one area.
Different problems,
shared questions.
and make better decisions?
Selected papers
Explore all publications & patents
Along the way
WorkMicrosoft AISenior Applied ScientistApr 2025 – Present
I work on foundation models for advertising.
Amazon AdvertisingApplied Scientist IIDec 2022 – Mar 2025
I worked on foundation models and multilingual models for ad moderation. Active learning reduced annotation needs by 80%. Fine-tuning with human feedback automated 16% more ad policy decisions.
TCS Research · MumbaiResearcher, Technical LeadMay 2018 – Dec 2022
I worked on supply chain optimization with deep RL (6 publications and 3 patents), multi-agent games including Pommerman (3 publications), and sample-efficient RL, general value functions, and policy optimization (4 publications).
TCS Research · Mumbai
TCS Research · GurgaonResearcherJul 2016 – May 2018
I studied sentiment analysis, stock market prediction from news, and behavior in Reddit and Twitter communities.
TCS Research · Gurgaon
Let’s talk research.
If you’re working on a related problem or have a question about a paper, feel free to email me.
hardik.meisheri@gmail.com