Nachiketa Gargi

Computer Science student from the SF Bay Area at the University of Michigan with a strong background in software development and audio production. Current interests include virtual/augmented reality, computer security, and audio signal processing.

About Me

I'm a sophomore passionate about computer science and music technology. Past projects have involved AR/VR, web design, digital signal processing, data science and machine learning, computer vision, and robotics. I have over 5 years of programming experience and I am also an electronic music producer with releases on several indie labels. My resume is available here.

  • University of Michigan '22 - B.S.E. Computer Science
    • Organizations: Project Music (Software Team Manager), WolverineSec, StartUM, UM::Autonomy
  • The Nueva School '18
    • Organizations: Programming Club, FRC Team 4904, Jazz Ensemble
Work Experience
  • Research Assistant - MIDAS Music Theory
    Sept 2018 - Present
    • Research assistant for the Michigan Institute for Data Science project "A Computational Study of Patterned Melodic Structures Across Musical Cultures," a collaborative research project between EECS, Math, and Music faculty
    • Developed novel method of optical mark recognition to detect and process Devanagari script and additional markings to produce a computer-readable archive of Indian music compositions documented in 1860. Corpus will be used for data analysis to determine data-driven musicological conclusions between other musical traditions including Irish folk music and Baroque music.
  • Full Stack Engineer - YouSound
    June 2017 - Sept 2018
    • Singlehandedly developed a scaleable backend for an artist chat platform using Node.js and, deployed at production scale using AWS.
    • Implemented a web-scale video live-streaming platform (like using AWS Elemental MediaLive, MediaPackage, and CloudFront.
    • Worked directly with founder and designers to implement and integrate Vue.js front-end for the chat platform using existing and new APIs while rapidly adapting to and using a consistent code style.
  • Software Engineer Intern - Primity Bio
    June 2016 - Aug 2016
    • Worked on a web-based realtime, collaborative data analysis platform for clients using testbased development with Node, Angular, and MongoDB.
    • This work was presented at an FDA conference in Washington, D.C.
  • U-M 2018 Music Makeathon - 1st Place
  • 2019 Winter StartUM Pitch Competition - 1st Place
  • 2019 Michigan Hackers Zero Day CTF - 1st Place
  • Dean's List
Programming LanguagesC/C++, JS, Python, Go, C#, Verilog
Frameworks and TechnologiesOpenCV, PCL, Unity3D, Unreal Engine 4, Arduino, Keras, Tensorflow, Numpy, JUCE, Vue
Creative SoftwareAbleton Live & Max/MSP, Blender, After Effects, Photoshop, Logic Pro, TouchDesigner
ToolsGhidra, Wireshark, Quartus, Git, AWS

Adversarial Reinforcement Learning for Music Generation
ISMIR 2018 Late-Breaking Session

  • Project Music - Software Team Manager
    Jan 2019 - Present

    Software Team Manager responsible for maintaining the club's online code presence, managing the software aspects of current projects, and participating in weekly meetings with the rest of the club's executive board.

  • Inviso - AR Spatial Audio Authoring Tool
    Sept 2019 - Present

    Working with Professor Anıl Çamcı on porting his WebGL-based Inviso spatial audio authoring tool to AR platforms including ARKit and ARCore. Prototyping AR interactions for an audio-focused application.

  • Music Makeathon - Realtime Sound-controlled Audio and Video Resampling
    Oct 2018

    Built a live, real-time audio processor with C++ and JUCE that uses an FFT comparison heuristic to replace input audio with audio from existing songs, producing a unique "remixing" effect. In addition, used Max to control video playback based on frequency and amplitude of input audio. The entire project was completed within 18 hours. Won first place in the U-M Project Music 2018 Makeathon. Featured in U-M Engineering Newsletter.

  • FieldAC - FRC Robot and Object Localization
    Jan 2018 - Apr 2018

    As part of FRC team, trained a custom model for an object detection framework (YOLOv3) on game pieces. Used a novel method for sensor fusion to maintain a field model using optical flow in conjunction with YOLOv3, onboard LiDAR, and IMU to estimate pose of robot and game pieces on the field. Model was used for an autonomous routine to manipulate the nearest game piece.

  • Kanooli - Electronic Music Production
    Oct 2016 - Present

    Regularly produce electronic compositions under the alias Kanooli. Developed branding strategy and accompanying website, and have accumulated over 200k plays through releases on several indie music labels with large audiences.

  • Schedulizer - School Assistant Bot for Telegram
    Aug 2017 - June 2018

    Developed Telegram chat bot used by the majority of students at my high school to assist students with the frequently changing class schedule and upcoming homework assignments.

  • AutoMuse - Automated computer music composition
    Oct 2014 - June 2016

    Used markov chains and LSTM neural networks to generate music from a dataset of scraped MIDI files.