About Me

Hi! I am currently pursuing my MS in Computer Science at Columbia. I graduated from UC Berkeley in 2023 with my BS in Electrical Engineering and Computer Sciences (EECS) with a Minor in Data Science. This website contains information about my research projects, personal projects, teaching experiences, and industry experiences.

Email:     pranav.sukumar@columbia.edu
Github:     https://github.com/Pranav-Sukumar
LinkedIn:     https://www.linkedin.com/in/pranav-sukumar/

Research

Columbia Computer Graphics and User Interfaces (CGUI) Lab

Graduate Researcher
(Aug 2023 - Present)

  • Developed a VR application for infant lumbar puncture training that provides procedure visualizations and feedback for trainees.
  • Visual Guidance for Infant Lumbar Puncture Training in XR Research Demo presented at conference and Extended Abstract published in conference proceedings.

UC Berkeley Game Theory (GamesCrafters) Lab

Undergraduate Researcher
(Aug 2022 - May 2023)

  • Worked on a project that combines augmented reality and game theory to aid in optimal play of a physical board game.
  • Implemented board game detection, built localization of the board game, and trained a neural network to detect game pieces.

UC Berkeley Algorithms and Computing for Education (ACE) Lab

Undergraduate Researcher
(Aug 2021 - May 2022)

University of Washington Personal Robotics Lab

Undergraduate Researcher
(May 2020 - Dec 2020)

  • Worked with the Assistive Dexterous Arm robot that performs assisted feeding for people with upper-extremity impairments.
  • Wrote a Hidden Markov Model (HMM) to predict when a person needs to be fed by the robotic arm.
  • Used computer vision algorithms to extract features for the HMM including body pose and facial expressions.

Personal Projects

MATER

MATER is a robotics research project that explores Autonomous Vehicular Pick-and-Place in an Unknown Environment. We built a custom gripper onto an existing RC car and implemented sensing, planning, and actuation algorithms so that the car can drive around an unknown room until it finds a tennis ball, then pick it up and bring it back to where it started. We implemented a PID controller, object detection, random exploration, and code to open and close our custom 3D Printed Gripper.

The MATER project has its own website, explaining the project in much greater depth. Please check it out!

Sign-ify

The iOS app Sign-ify, developed for CalHacks 6.0, enables enhanced live-lecture for members of the hearing-impaired community by intelligently converting a professor's speech to a sequence of American Sign Language (ASL) videos for the user to watch during lecture. There is a dissonance in grammatical structure and syntax between written language and ASL (reading in english rather than ASL is akin to reading in a non-native language). Sign-ify is able to translate professors’ dictation to pre-recorded ASL videos in near-real time, eliminating the need for students to interpret written English and having to “translate it” into ASL as they process information.

Sign-ify was an award winning project at CalHacks 6.0! Please check out the DevPost for more information!

Edify

Edify allows Professors and TAs to gather insights about class engagement and questions from a recorded Zoom meeting. This application takes in the Zoom Video recording, Zoom chat transcript, and Zoom audio transcript, which can be generated from a recorded meeting. We use computer vision to analyze how often the students are looking at the computer screen and paying attention. We also use Natural Language Processing (NLP) to provide sentiment analysis and topic clustering for the questions posed in the audio transcript and chat box. We then email these charts and statistics to an email address passed in to the Web-application.

RecipeBot

RecipeBot is a chat-bot compatible with Google Assistant that recommends recipes based upon spare ingredients a user inputs. The bot has been trained on more than 1000 ingredients.





GradeScoper

GradeScoper is a chrome extension that scrapes Gradescope for assignment deadlines and adds them to a Google Calendar.

Speech Matrix Solver

Speech Matrix Solver is a web application our team built in order to solve complex systems of equations by performing Gaussian Elimination on an augmented matrix verbally dictated by the user. Both the Gaussian Elimination backend and frontend were built from scratch using JavaScript and HTML5.

Teaching

Teaching Assistant for Database Systems

Columbia COMS 4111
(Aug 2023 - Dec 2023)

  • Course taught by Professor Luis Gravano.
  • Debug student code in office hours for project implementing a database application with a web front-end.
  • Created and maintained the class PostgreSQL server running in a cloud VM.

Teaching Assistant for Database Systems

UC Berkeley CS 186
(Aug 2022 - May 2023)

  • Course taught by Professor Alvin Cheung.
  • Teach weekly sections, explaining database management system theory and algorithms from class including B+ Trees, Buffer Management, Query Optimization, Recovery, Parallel Query Processing, Distributed Transactions, and Distributed Systems.
  • Debug student code in office hours for Java Project implementing a Database Management System with a B+ tree index, System R query optimizer, concurrency, and an ARIES recovery manager.
  • Edit course notes and textbook by adding new content presented in lecture and example problems with solutions.

Reader for Computer Security

UC Berkeley CS 161
(Jan 2022 - May 2022)

  • Course taught by Professor Nicholas Weaver.
  • Course topics include: Memory Safety Vulnerabilities, Cryptography, Block Ciphers, Hashes, MACs, Certificates, CSRF, SQL Injection, Networking, DNSSEC, Firewalls, Blockchain.
  • Reviewed student design documents for a project building an End-to-End Encrypted File Sharing System in Golang.
  • Wrote and graded exam and homework questions.

Computer Science Mentors at Berkeley

Mentor for UC Berkeley Data Structures (CS 61B)
(Aug 2020 - May 2021)

  • Course taught by Professor Paul Hilfinger.
  • Taught small sections of five students, reinforcing data structures and algorithms concepts like Object Oriented Programming, Runtime Analysis, Hashing, Trees, Shortest Path, Graph Traversal, Heaps, and Sorts.

UC Berkeley Data Structures (CS 61B) Academic Intern

(Aug 2020 - Dec 2020)

  • Course taught by Professor Paul Hilfinger.
  • Helped students debug Java projects and homework in labs.

Work Experience

Databricks

Software Engineer Intern
(May 2024 - Aug 2024)

Summer internship with Databricks (Delta Sharing Team):

  • Worked on Delta Sharing which is a part of the Delta Lake foundation.
  • Added support to the open-source Delta Sharing server and client-side connector to support sharing tables with Deletion Vector and Column Mapping features.
  • Added limit and JSON predicate hint pushdown on the server to optimize query processing.
  • This was requested by 250+ Databricks customers and my work can be found on the Delta Sharing open source repository.
  • Prototyped cloud token sharing in open-source Delta Sharing client-side connector as a stretch project.

Amazon Robotics

Return Software Engineer Intern
(May 2023 - Aug 2023)

Summer internship with Amazon Robotics:

  • Expanded my previous internship project by connecting augmented reality device to the PC controlling production workflow.
  • Built a system that uses DynamoDB to securely send and receive data between the device and Fulfillment Center Station PCs.
  • Networking system was extensible and scalable allowing it to be utilized in other projects and parts of the production workflow.

Amazon Robotics

Software Engineer Intern
(May 2022 - Aug 2022)

Summer internship with Amazon Robotics:

  • Built an Augmented Reality application for the Microsoft HoloLens 2 to replace current infrastructure in a robotic workcell.
  • The augmented reality application replaced a complex network of proprietary cameras, lighting devices, and overhead canopies.
  • Research project that involved structuring and defining the research problem from a one-line description, outlining goals for the project, and defining measures of success.
  • Developed workcell localization, interaction detection, and a user interface in Augmented Reality using C#.
  • Applied for a design patent; This project saves ~ $1 million per Amazon Delivery Center.

Apple

AI/ML Intern
(Jan 2022 - Apr 2022)

Spring internship with the Siri Response Framework Team:

  • Authored multi-thousand-line Swift programs showcasing the complete set of new APIs for use by Apple internal developers.
  • Addressed enhancements and bugs in Siri Response Framework as a part of iOS 16.0.

Nvidia

Returning Software Engineer Intern
(Jun 2021 - August 2021)

Summer internship with the Nvidia Cloud Infrastructure & Platform team:

  • Developed secure end-to-end automation of Windows KMS Volume Activation for NVIDIA GeForce Now machines.
  • Used Terraform and Python to automate the creation of AWS EC2 instances, VPCs, Subnets, and Security Groups.
  • Developed a Jenkins Pipeline to build, deploy, and test the Windows KMS Volume Activation automation.

Nvidia

Software Engineer Intern
(May 2020 - August 2020)

Summer internship with the Nvidia Cloud Infrastructure & Platform team:

  • Improved monitoring capabilities for cloud services by writing Golang programs to expose CPU performance and event metrics.
  • Scraped exposed metrics for Prometheus to build visual Grafana dashboards for the NVIDIA GPU Network platform.
  • Tested and deployed code on Docker containers running under Kubernetes on virtual machines in a Linux environment.
Recieved and accepted a return internship offer.

Expedia

Software Development Intern
(June 2018 - August 2018)

As part of the Expedia Search and Suggest Team, I worked on both front-end and back-end capabilities to improve Expedia's homepage search experience. I used Java, JavaScript, and HTML/CSS to develop these features:

  1. Categorizing the types of suggestions displayed so that in inline search suggestions, the user can distinguish between SEA as a city abbreviation and SEA as an airport code
  2. A customer feedback tool for the quality of search results
  3. Eliminating the unnecessary need for the user to explicitly grant location access to Expedia in the current location feature by creating a heuristic based current location option
About 700,000 customers per week interact with the features I built. I also participated in Expedia Hackathon 7.0 and had 4th most popular project.