Jul 2021 – Present · 4 yrs 11 mos · Chennai, Tamil Nadu, India · Hybrid

• [Research] Machine Learning assisted discovery of high entropy alloys (HEAs) for electrocatalytic energy conversion applications. HEAs are alloys made up of 4 or more elements and have shown enhanced properties for various use-cases. Screening such novel material involves solving complex schrodinger equations which are highly compute intensive and is challenging task and thus creating a bottleneck. In this endeavor we aim to use machine learning to simplify some this processes in the material screening pipeline. (In collaboration with IITB)
• [Applied] Automated the manual work of a team of application engineers maintaining the xapth configs by a custom algorithm that dynamically generate xpaths. Xpaths are used to extract critical information from web pages and thus enabling continuous monitoring. Authoring xpath is usually equivalent to writing regex to find an element uniquely in a HTML, which usually takes 2-5 days for an engineer. The devised algorithm uses concepts from graph theory in conjunction with genAi to build these xpaths thus reducing human in the loop. #patent #paper #AMLC
• [Applied] Built a product matching model for the Prime Video business using LLMs. The LLM was fine tuned on general corpora instead of domain specific data to achieve better results. Published this finding in internal ML conference. #paper #AMLC