Skip to content
View Pranjulcr7's full-sized avatar
🤖
🤖

Highlights

  • Pro

Block or report Pranjulcr7

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Pranjulcr7/README.md

Pranjul Gupta: Software and ML systems engineer. Ex-Disney+. MS candidate, Carnegie Mellon.

pranjulgupta.com Ask the assistant LinkedIn Hugging Face Résumé Email

Backend and ML systems engineer. Four years at Disney+ (Hotstar, India streaming) owning Continue Watching across phone, TV, and web, and cutting post-match traffic 72.5%, from 8 million requests a minute to 2.2 million. Finishing an MS in AI Engineering at Carnegie Mellon. Open to senior software and AI/ML roles from January 2027.

Open source: Spanish ASR recipe for ESPnet, merged (pull request #6491 · model on Hugging Face)

Pittsburgh, PA, open to relocation · Work authorization: STEM OPT (3 years)

Career numbers: see the website for each number's scope

Selected work

Work What it is Result
Spanish ASR recipe for ESPnet
WAVLab, Carnegie Mellon University
Open-source ML / Evaluation and debugging
A Spanish speech recipe for ESPnet, and the evaluation bug that was too good to be true. Dev WER 24.4%; overall test WER 56.3% under substantial domain and channel shift; 18.9% WER on the 711-utterance spontaneous-answer subset with novel speakers.
PR · model
Continue Watching
Disney+
Distributed systems / End-to-end ownership
The tray that remembers where you stopped, for tens of millions of viewers. 72.5% fewer home-page requests per minute after a live match (8 million down to 2.2 million in that post-match window).
Two-phase tray ranking
Disney+
Production ML / Low-latency serving
Ranking 200+ trays per session in 50 ms at p99. Ranking service p99 latency of 50 ms across 200+ candidate trays per session (service latency, not end-to-end page latency).
pranjulgupta.com
Personal project
An assistant that can only say what the profile says. 32 evaluation questions check retrieval and answers, including facts that are not on record (my age, Rust) and prompt-injection attempts, which must not produce an invented fact or the system prompt.
live · ask it

Live from the assistant

The assistant on my site answers only from my published profile and shows which sections it used.

Live numbers from the portfolio assistant

Refreshed daily by a GitHub Action from the site's own logs (Cloudflare D1). Site telemetry, not career statistics.

Stack


Car render in the banner adapted from (FREE) Porsche 911 Carrera 4S by Karol Miklas, CC BY-SA 4.0. Not affiliated with or endorsed by Porsche. Everything on this page is generated from the same profile file as pranjulgupta.com.

Popular repositories Loading

  1. College-android-chatbot College-android-chatbot Public

    College app with attendance management system( where faculty can take and check student attendance, and student can login to see their attendance), feedback to give feedback about the app and integ…

    Java 9 7

  2. tensorflow-projects tensorflow-projects Public

    Forked from edyoda/tensorflow-projects

    Jupyter Notebook 1

  3. Data-Science--Cheat-Sheet Data-Science--Cheat-Sheet Public

    Forked from georgearun/Data-Science--Cheat-Sheet

    Cheat Sheets

    2

  4. handson-ml2 handson-ml2 Public

    Forked from ageron/handson-ml2

    A series of Jupyter notebooks that walk you through the fundamentals of Machine Learning and Deep Learning in Python using Scikit-Learn, Keras and TensorFlow 2.

    Jupyter Notebook

  5. King-County-House-Sales-Prediction King-County-House-Sales-Prediction Public

    Jupyter Notebook

  6. Sachin_Tendulkar_ODI_EDA Sachin_Tendulkar_ODI_EDA Public

    Exploratory data analysis of Sachin Tendulkar's performance in ODI.

    Jupyter Notebook 1