Vedarsh Shah

Computer Vision Team Lead at Duke Robotics Club
  • Claim this Profile
Contact Information
us****@****om
(386) 825-5501
Location
Durham, North Carolina, United States, US
Languages
  • Spanish Limited working proficiency
  • Gujarati Native or bilingual proficiency
  • Hindi Limited working proficiency
  • English Native or bilingual proficiency

Topline Score

Topline score feature will be out soon.

Bio

Generated by
Topline AI

5.0

/5.0
/ Based on 1 ratings
  • (1)
  • (0)
  • (0)
  • (0)
  • (0)

Filter reviews by:

Corey McMillan

I was a mentor to Vedarsh and his team during Optum's 2022 Technology Development Program summer internship and during that time he demonstrated excellent team leadership and collaboration, project management, stakeholder communication, and technical skills. From the very beginning of the engagement he was an integral part of the team, setting clear scope boundaries as to not overwork the team as well as providing clear documentation of the code to assist in the hand-off of the project once the internship was finished.

You need to have a working account to view this content.
Join now
You need to have a working account to view this content.
Join now

Credentials

  • Microsoft Office Specialist: Master 2016
    Microsoft
    Nov, 2020
    - Oct, 2024
  • RCM Level 10 Certified (Piano)
    The Royal Conservatory of Music
    Aug, 2020
    - Oct, 2024

Experience

  • Duke Robotics Club
    • United States
    • Research Services
    • 1 - 100 Employee
    • Computer Vision Team Lead
      • Aug 2022 - Present

      • Leading Duke Robotics Computer Vision team of 10, with measurable goals, detailed planning, and effective execution.• Scored 100% on all computer vision tasks undertaken at RoboSub 2023 Autonomous Underwater Vehicle competition.• Created work breakdown structure of 115+ tasks, delegated, and empowered that transformed into high performing team.• Engineered a computer vision pipeline using Python, Robot Operating System (ROS), and Docker containers.• Achieved 99% mean average precision on the YOLOv7-tiny model and increased detection rate from 1Hz to 22Hz.• Designed an underwater simulation in Unity 3D, generating 10,000+ auto-labeled training images.• Partnered with mechanical team to design waterproof enclosure ensuring underwater camera vision with <2% distortion.• Collaborated with electrical team to troubleshoot estimation of robot position and depth, achieving 70%+ accuracy.• Built GUI using PyQt that displayed real-time camera feed with detections and enabled live monitoring of parameters. Show less

    • Computer Vision Team Member
      • Aug 2021 - Aug 2022

      • Performed YOLOv4 object detection and stereo depth perception using DepthAI and OAK-D-PoE camera• Authored automated scripting to evaluate 16 image preprocessing methods for performance and image quality• Developed Python scripts to extract and sync image feeds from rosbag files, and computed depth map using stereo vision• Won Best Technical Report at RoboSub 2022

  • Fidelity Investments
    • United States
    • Financial Services
    • 700 & Above Employee
    • Software Engineer Intern
      • Jun 2023 - Aug 2023

      • Delivered voice-enabled conversational stock screener leveraging NLP/AI technology that transformed client engagement. • Achieved reduction in ask-to-results time from 3 min to under 30 sec, empowering quicker and informed investment decisions. • Invented a custom JSON query language and designed OpenAI prompt that accurately extracted 150+ filters from voice queries. • Developed a robust backend using JavaScript and Node.js, ensuring the system’s efficiency and performance. • Integrated Azure Speech to Text API and Azure OpenAI GPT-3.5 API that transcribed audio and recognized 20+ complex entities. • Designed an intuitive, zero-training mobile and desktop interface using Figma ensuring swift user adoption. Show less

  • Optum
    • United States
    • Hospitals and Health Care
    • 700 & Above Employee
    • Software Engineer Intern
      • Jun 2022 - Aug 2022

      • Engineered an innovative web app for continuous engagement of behavioral health clients that improved treatment outcomes. • Saved 40%+ health professionals’ appointment time from collecting client updates to treatment formulation. • Led a team of 5 for full-stack development, integration & testing using Agile methodology, resulting in on-time project delivery. • Built adaptive topological graph of client psyche using JavaScript, React, Python, Flask & Elasticsearch, delivering robust system. • Deployed OpenAI GTP-3 API for summarization of client journal entries, enhancing the productivity of the health professionals. Show less

  • UbiStrap
    • Durham, North Carolina, United States
    • iOS Engineer
      • Mar 2022 - May 2022

      • Developed 10 core app features using Swift, including home, profile, workout, and activity tracker pages, delivering intuitive UX. • Empowered users to effortlessly access health and fitness data, increasing app usage by 35%. • Developed 10 core app features using Swift, including home, profile, workout, and activity tracker pages, delivering intuitive UX. • Empowered users to effortlessly access health and fitness data, increasing app usage by 35%.

  • University of Cincinnati
    • United States
    • Higher Education
    • 700 & Above Employee
    • Machine Learning Research Intern
      • Jun 2021 - Aug 2021

      • Employed machine learning technique of Latent Dirichlet Allocation (LDA) topic modeling that classified 76 HIPPA breach norms. • First author on research paper “Towards Norm Classification: An Initial Analysis of HIPPA Breaches” published in ESPRE 2021 • Employed machine learning technique of Latent Dirichlet Allocation (LDA) topic modeling that classified 76 HIPPA breach norms. • First author on research paper “Towards Norm Classification: An Initial Analysis of HIPPA Breaches” published in ESPRE 2021

  • University of Cincinnati
    • United States
    • Higher Education
    • 700 & Above Employee
    • Research Intern
      • Jun 2020 - Aug 2020

      ▪ Analyzed 40,000+ lines of Java code and created an algorithm to extract norms from code assessing regulatory compliance ▪ Developed ontology of FERPA using Protégé to measure gaps between reported FERPA breaches and regulatory coverage ▪ Analyzed 40,000+ lines of Java code and created an algorithm to extract norms from code assessing regulatory compliance ▪ Developed ontology of FERPA using Protégé to measure gaps between reported FERPA breaches and regulatory coverage

  • SchoolAppsUSA
    • Mason, Ohio
    • Software Developer Intern
      • Jan 2020 - Aug 2020

      • Developed onboarding module, home page, and navigation infrastructure of How2Life mobile app using Flutter/Dart • Collaborated with a globally distributed development team • App engages students ages 13-18 and provides educational content and resources for improving mental health and wellness • App used in 6 schools with 900+ students, having 72% monthly active users and 93% positive reviews on mental health contents • Developed onboarding module, home page, and navigation infrastructure of How2Life mobile app using Flutter/Dart • Collaborated with a globally distributed development team • App engages students ages 13-18 and provides educational content and resources for improving mental health and wellness • App used in 6 schools with 900+ students, having 72% monthly active users and 93% positive reviews on mental health contents

Education

  • Duke University
    Bachelor of Science - BS, Computer Science and Statistics
    2021 - 2025
  • William Mason High School
    High School Diploma
    2017 - 2021

Community

You need to have a working account to view this content. Click here to join now