AdSkate is an ad-tech firm building solutions that improve the effectiveness of digital advertising campaigns. We are developing AI tools that extract hidden insights from advertisers' image and video creative assets to help them create content that stands out in today's content-saturated world. Our clients, including DSPs, platforms, brands, and agencies, leverage our data to execute high-performing digital campaigns.AdSkate's AI enables advertisers to discover the best performing design elements within their image and video creatives. Creatives get converted into data points such as - imagery, detected activities, visual themes, brand logo positioning, verbal ad messaging, visual ad messaging, verbal call to action, visual call to action etc. Our analytics pipeline then correlates these data points with campaign performance metrics. This allows our clients to design creatives with the elements that will drive up their intended metric whether that be clicks, view rates, video playback rates, number of app installs etc.AdSkate’s ‘Creative Intelligence’ solution has enabled advertisers across industries to increase click through rates by 75 per cent and view rates by 40% across multiple channels ranging from display, video to social media. To learn more, visit https://www.adskate.com/

- Conceptualizing, designing, developing multi sensor solutions to improve overall imaging quality on robot imaging stack, in keeping with agreed upon customer SLAs- Formulating quantitative metrics for automated remote image quality analysis to alert robot operations team in real time of potential issues on the field- Exploring and adapting open source image processing and computer vision software tools to supplement existing algorithms in the AI and vision software stack

Design and development of video analytic and computer vision solutions for embedded and cloud platforms with a specific focus on smart home environments. I am specifically in charge of- Conceptualizing and designing human action recognition framework deployed on embedded platforms of Hawxeye's customers- Managing Hawxeye's cloud based analytics team to develop and release our GPU based analytics toolkit encompassing deep learning based solutions for facial recognition and object detection- Interacting with customers to help integrate our framework into their product development pipelines- Framework and API customizations to enhance utility on customer side- Containerizing our software evaluation kits to enhance portability across customer machines- Working with patent lawyers for timely filing provisional and full patents of relevant IP

- Worked on user authentication using Radio Frequency Impedance Interrogation based cardio-synchronous waveforms- My research here involved the investigation of the utility of wavelet based features and sub-space modeling for user authentication, using this biomedical signal modality- I developed a dictionary based classification tool that utilized signal sparsity for this project- Results based on my research have been submitted to IEEE Transactions in Biomedical Engineering

- Worked as a research intern prior to joining the Phd programme at Carnegie Mellon University- Designed and developed a commercial PTZ camera based user tracking and iris recognition system- Analyzed the use of various local and global optimization techniques, in choosing an optimal Gabor filter bank for iris recognition