Making the physical world intelligent

I've always been fascinated by the idea of making machines intelligent. I am building efficient algorithms, which are intuitively similar to the selective attention mechanism of the human brain. I'm in the 4th year of my PhD at Penn State and am the Founder at Lightscline, where we are building embedded AI to eliminate up to 90% of sensor data analytics costs by teaching machines to focus on the 10% important data. Here's my story in four short paragraphs:

1. Last fall, I was awarded the NSF I-Corps National Teams award ($50k) for customer discovery related to my deep tech startup. I've done 220+ customer meetings, visited and presented at 6 tech expos, and am working on 8 pilots now.

Some recent news: https://www.psu.edu/news/invent-penn-state/story/penn-state-research-based-startup-completes-nsf-i-corps-national-teams/

NSF I-Corps National Teams video:

A short product demo:

2. There's going to be 73+ Trillion GB of sensor data by 2025. Contemporary techniques are not designed to work on this scale of data. In my research, I'm working on AI-techniques to collect 2-10x fewer data, inspired by mechanisms such as selective attention used by the human brain. Our paper describes the problem: https://par.nsf.gov/servlets/purl/10288794

3. As a PhD student at Penn State and in my prior work, I've done several interesting experiments as I like to work at the intersection of hardware and software. Here are a couple of videos from my Youtube channel: https://www.youtube.com/@Ankur_explores

4. I've added a pictorial story of my manufacturing journey prior to 2020 at: https://ankurashutosh.wixsite.com/ankurverma/work

Includes:

• My travel to different manufacturing facilities in India and Germany.

• Some personal projects!