-Joined Verseon as a part of the Edammo acquisition.-Leading a team developing breakthrough applications of machine learning in healthcare/pharmaceuticals.

Teaching classes in Computer Information Systems.

-Edammo was acquired by Verseon in 2022.-New approach to commercial AI applications.-Edammo chosen as one of Top 10 Predictive Analytics providers for 2021 by CIO Applications.-Edammo chosen as one of Top 10 AI solution providers for 2019 by PharmaTech Outlook.-Let the software engineering/data science team.-Delivered cloud-based solution to multiple customers-Primary architecht of the software solution

Consulting in the fields of machine learning, image/video analysis, video streaming, and cloud services architecture.

- Breakthrough video compression technology applicable to existing codecs including H.264 and H.265. Based on a proprietary object analysis paradigm.- Lead inventor on 23 filed patents of which 4 have issued to date.- Lyrical Labs was chosen for the Cable Labs Innovation Showcase in 2012.- In an independent testing by Cable Labs in 2014, Lyrical Labs H.264 encoder showed an improvement over x264(a leading H.264 implementation) by over 30% for live 720p encoding.- Architected the Lyrical Labs Cloud Encoding Service that was based in AWS. Leveraged EC2, S3 and SQS for dynamic scaling and asynchronous work load distribution.- Recruited and hired the engineering team.- Lyrical Labs encoding powered the launch of The Great Courses site(www.thegreatcourses.com) and Hoichoi (www.hoichoi.tv) an Indian Netflix-like destination site.- In independent testing at Encompass and NBC Advanced Media Labs(2017), Lyrical Labs H.264 encoder showed over a 50% improvement over leading commercial H.264 encoders for 1080p live content.- Published 6 papers in academic peer-reviewed literature.- Point of contact for customer relationships responsible for day-to-day customer relationship management.For an example of the research, see:https://research.nvidia.com/publication/adaptive-segmentation-based-learned-quality-metric

Taught classes in research areas of interest: Object Recognition, Image and Video Compression, etc.

Co-founder. Breakthrough algorithms for automated tracking, extraction and recognition of objects in complex video scenes. We leveraged our core technology to bring a new paradigm to video advertising on the Web.Product - Robust in-video ad insertion product based on core video analysis technology proven in real-world deployment. Successfully analyzed almost 1M different video clips in an unsupervised deployment. Product was able to run with no manual intervention. Technology - Co-inventor on 8 patents filed (2 issued to date). Wrote initial versions of the segmentation and segment motion estimation modules. Co-managed all the algorithm development.Product Architecture - Created specs and architecture for all customer related product functionality.Team building - Recruited and hired engineering team.Entrepreneurship - Raised VC funding as a part of a 2-member founding team. Also, raised non-equity bridge from Comerica Bank.Business Development - Signed up the first 3 company customers.Customer Interaction - Primary technical point of contact for all company customers/partners and primary point of contact on all aspects of the relationship for 6 of 8 signed customers/partners. First deployed pilot resulted in customer putting out own press release about pilot success.

-Technical lead for off-shore team(Moscow) in developing new computational engine for a market-leading KLA-Tencor optical measurement product. Developed first time-domain broad-spectrum pseudo-spectral electromagnetic solver for dispersive materials. -Inventor/co-inventor on 4 issued patents and 3 filed patents. -Algorithms involve computational electromagnetism, image reconstruction and inverse problems.-Wrote proposal that was funded by Intel for a collaboration with Lawrence Livermore Lab for study of optical properties of nano-materials. Research resulted in 4 published articles.-Internal expert consultant for the E-beam modeling group.-Responsible in division for evaluating 3rd party algorithmic technology. Conducted an extensive evaluation of technology from a Boston based start-up.

Led a team in developing a proprietary object-based video codec. Including: -Created a new robust video segmentation technology. This technology has the capability to automatically find objects in a generic video scene. -Created robust motion estimation algorithms for objects. -Created algorithms for reconstructing images by motion compensating and layering objects. -Created algorithms efficiently coding uncovered areas in an image where no previous object existed. -Created algorithms for efficiently coding errors resulting from reconstructing images by applying object motion compensation and layering. -Wrote initial versions for 3 of 5 major algorithm modules(C/C++). Results indicated a 4:1 compression gain over the current MPEG2 standards, verified independently by teams from Thomson and Sony.Inventor/co-inventor on over 50 patents filed. 21 granted (all that have been examined so far).

-Developed novel linear circuit reduction techniques. The reduced representation was used directly in the ACE simulator engine of TimeMill, the market leading transistor-level circuit simulation product.-Obtained 95-99% reduction in circuit size versus the 65% that were obtained by previous reduction methods.-Work involved Krylov subspace methods, sparse matrix methods, eigenvalue decomposition, graph theoretical algorithms and novel matrix structures. Wrote all the code in C.

-Developed high-speed analog circuit simulator featuring very high-speed and accuracy. The accuracy is high enough to exhibit detailed electrical phenomena. Code written in combination of C/C++.-Obtained 20-100X speedups over SPICE while retaining SPICE level accuracy.-Work involved circuit partitioning, differential equation solvers and analog transistor device modeling.