S

Sek Chai

About

Detail

Co-Founder and CTO, Board of Director, Entrepreneur, Technology, Advisor
New York, United States

Timeline

work
Job
school
Education
auto_stories
Publication

Résumé

Jobs verified_user 0% verified
  • Latent AI
    Co-Founder and CTO, Board of Director, Entrepreneur, Technology, Advisor
    Latent AI
    Dec 2018 - Current (7 years 7 months)
    Latent AI is a US-based/owned startup spin-out of SRI International. Latent AI enables the creation of more robust, efficient, and agile edge AI applications. Our technology optimizes and secures AI model, while allowing for model adaptation and field updates. We have a team with proven startup experience (over +20 years in ML/AI and embedded software). I am responsible for company building, technology / product strategy, and customer success. I cover multiple roles from product development, engineering, sales to operations. I help build the federal business, with a growing pipeline. I now serve as a member of the board of directors.
  • V
    General Chair
    VIEW Vision Industry and Entrepreneur Workshop at CVPR
    Jun 2011 - Jun 2017 (6 years 1 month)
    I am the founder and general chair for VIEW - Vision Industry and Entrepreneur Workshop, with goals to bridge the gap between academia and industry in the computer vision. We held the first Venture Pitch contest at CVPR, and had a host of innovative talks ranging from startup funding, IP planning, and platforming. After 6 years of innovative programs, VIEW is now fully integrated with CVPR EXPO Spotlights.
  • SRI
    Technical Director, Information and Computing Sciences
    SRI
    Apr 2010 - May 2019 (9 years 2 months)
    Manage and develop technology portfolio for autonomous, mobile, low power, high performance computer vision and machine learning systems. Direct technology strategy and IP for embedded systems. Secure external funding from both government and commercial sources. Manage, recruit and hire top-notched best-in-class talent. Collaborate and team with academic and industry partners. Help run a profit and loss (P&L) business unit for 100 person lab. License IP and spin-off new ventures. Won and launched multi-million dollar research programs ranging from early-stage to mature technology development. Build and evangelize SRI embedded vision, computational imaging, and bio-inspired processing solutions. Performing deep learning consultancy for
  • E
    Steering Committee
    Embedded Vision Workshop at CVPRICCVECCV
    Jun 2008 - Aug 2020 (12 years 3 months)
    Proliferating "Embedded Vision" space as a key aspect of computer vision research. Champion the annual event to educate others of the technology. Bridge the gap between academic and industry. Book co-editor, Embedded Computer Vision. Journal guest editor, CVIU and JSPS Special issues on Embedded Vision.
  • C
    Owner and Co-founder, President
    ComputerVisionCentralcom
    Jun 2008 - Jul 2017 (9 years 2 months)
    I co-founded ComputerVisionCentral.com, a leading on-line community for computer vision and related technologies. It provided a source of daily news on computer vision; It was the number one listing for computer vision jobs; It was the host for the Embedded Vision Workshop (EVW) and the Vision Industry and Entrepreneur Workshop (VIEW). In its latest evolution, it served over 3000 registered users in a platform that is completely open and free.
  • Motorola Mobility a Lenovo Company
    Principal MTS
    Motorola Mobility a Lenovo Company
    Jan 2000 - Apr 2010 (10 years 4 months)
    Various technical positions, starting at individual contributor to team leader/mgr. * IP set top box development, media and content search, media transcode to mobile devices. Software developer role and program management * 3D GUI using M3G and OpenGL. Team mgr/lead role. * Touch-based interfaces for consumer electronics. IP discovery role. * Mobile Computer Vision. Search "chai mobile challenges" * IP video teleconferencing. Software developer role. * Camera & FPGA (Altera and Xilinx) design for image processing and lens correction. Architect & developer role. * Automatic license plate recognition. Search "Proven Force Multiplier". Architect and hardware developer roles. * ESL tools for hardware stream accelera
  • G
    Research & Teaching Assistant
    Georgia Tech
    Jan 1996 - Dec 2000 (5 years)
    * High performance, high throughput (systolic and SIMD) focal plane computing * Bi-directional fiber communication for automotive * Through-wafer optoelectronic interconnects for high-speed on-chip networks * Low power architectures for gigascale integration. MARCO & DARPA funding support.
  • IBM
    Coop / Intern
    IBM
    Jan 1992 - Jan 1994 (2 years 1 month)
    OSI-Lower Layer Conformance Test Center, Research Triangle Park, NC Multimedia Lab, (PC-TV product), Atlanta, GA
Education verified_user 0% verified
  • Georgia Institute of Technology
    Doctor of Philosophy (PhD, Electrical Engineering
    Georgia Institute of Technology
    Jan 1990 - Dec 1999 (10 years)
    Sek earned his BSEE, MSEE, & Ph.D in Electrical Engineering from Georgia Institute of Technology. His doctoral research, "Real Time Image Processing on Parallel Arrays for Gigascale Integration", addresses high throughput, efficient architectures such as systolic arrays, and the different ways to map data flow graphs onto them within the constraints of on-chip interconnect and power consumption. Advisor: Dr. D. Scott Wills. Co-advisor: Dr. Jim Meindl
Projects (professional or personal) verified_user 0% verified
  • I
    IARPA Thor
    Leading a team as system architect to build a system to detect biometric presentation attacks (e.g. to counter those trying to fool fingerprint, iris and face recognition systems). We are using deep temporal models to detect and recognize these attacks. Our approach uses a unique technique on Dynamic Biometrics. Our research will deliver a biometric prototype that is more robust to attacks with advances in sensor processing, fusion, and anomaly detection.
  • D
    DARPA Cortical Processor
    There many types of temporal data that we encounter in our everyday life (audio, video, gesture) and deep learning models that support efficient time-series analysis will support higher classification performance. I am a lead PI on the DARPA Cortical Processor. My team includes Graham Taylor (U.Guelph) and Roland Memisevic/Yoshua Bengio (U.Montreal). We are exploring models (Restricted Boltzman Machine - RBM, RNNs/LSTM) for multi–modal data. Our research is focused on how and when to fuse the data and learnt concepts (e.g. early/late fusion).
  • D
    DARPA ADAPT
    Lead team to develop new compression-and-search technology for mobile smartphone processors. Enable full-motion video (FMV) at bit rates up to 10 times lower than rates typically used in commercial broadcast applications—allowing users to search, tag and share video in bandwidth-limited areas.
  • L
    Lifelong Learning Machines
    We are developing a next-generation AI systems that applies continuous learning to become better and more reliable at performing new tasks. Our approach uses generative memories that are coupled reinforcement learning. We demonstrate results on StarCraft2 strategy game simulations, where we interject surprises to measure AI responses to catastrophic forgetting.
  • I
    IARPA RAVEN
    Developing a micro data-center and new algorithms for imaging sub 10nm integrated circuits for semiconductor process verification and failure analysis.
  • L
    Low Power Computer Vision Accelerators
    Help build and grow new business based on licensing computer vision IP (FPGA, GPU, smartphone). Lead team to mature architecture framework that leverage technology locked in an proprietary ASIC. Develop go-to-market strategies for drone, handheld and surveillance applications (tracking, visual enhancements, augmented reality).
Publications verified_user 0% verified
  • Springer
    Efficient Object Detection Using Embedded Binarized Neural Networks
    Springer
    Jan 2017
  • A
    Neurocube: A Programmable Digital Neuromorphic Architecture with High-Density 3D Memory
    ACMIEEE International Symposium on Computer Architecture ISCA
    Jun 2016
  • G
    NeuroCube: A Scalable, Efficient, Platform for Neuro-Inspired Computing
    Government Microcircuit Applications Critical Technology Conference GOMACTECH
    Mar 2016
  • I
    Enabling Smart Camera Networks with Smartphone Processors
    IEEE Computer Magazine
    Jan 2015
  • Springer
    Designing Vision Systems that See Better, Advances in Embedded Computer Vision, Part III
    Springer
    Jan 2014
  • P
    Forensic prescreening system using coded aperture snapshot spectral imager
    Proc of SPIE Vol H
    Jan 2014
  • Springer
    Embedded Computer Vision
    Springer
    Jan 2008
  • I
    An Optimized, Programmable In-Memory Accelerator for Deep Learning
    International Symposium on Computer Architecture ISCA Submitted
This is a community-created genome.