Profile banner
Sri Raghu Malireddi

Sri Raghu Malireddi

Machine learning engineer and founder — on‑device ML, CV, and ML systems

sri
San Francisco Bay Area
Joined October 2025
Favicon imagesriraghu.com
Favicon imagegithub.com
Favicon imagescholar.google.com
Favicon imagekaggle.com
Favicon imagestackoverflow.com
Favicon imagex.com
Favicon imagelinkedin.com

Network

3.8K connections

Summary

On‑device ML and model optimization practitioner: Sri Raghu has publicly presented and written about post‑training optimization, quantization, and deploying models on device. His recent industry role at Grammarly is explicitly on on‑device ML and LLM optimization, and he has spoken at industry events on these topics. analyticsvidhya+1
Research and dataset/benchmark author in computer vision: His MSc thesis from University of Victoria focuses on systematic dataset generation and benchmarks for modern computer vision (including hand segmentation and local‑feature benchmarks). He is a co‑author on the HandSeg dataset/papers and several conference papers arising from his graduate work. uvic+1
Experienced software engineer with applied research outputs: He spent multiple years at Microsoft (2018–2023), is listed on patent applications assigned to Microsoft, and maintains practical engineering artifacts (GitHub repos) demonstrating mobile and ML system implementations. openreview+2
Teacher, mentor, and community contributor: He has taught/mentored (BrainStation, Udacity), contributed on StackOverflow and Kaggle, and maintains technical writing on Medium/personal blog — indicating a sustained interest in education and community knowledge sharing. kaggle+2

Work

Education

Projects

Writing

LITE: Intent-based Task Representation Learning Using Weak Supervision (NAACL 2022 proceedings)

January 1, 2022

A paper on intent/task representation learning using weak supervision co‑authored with researchers including Oriana Riva and others (appeared in NAACL/ACL venues).

Favicon imageaclanthology.org

Systematic Generation of Datasets and Benchmarks for Modern Computer Vision (MSc thesis)

January 1, 2019

MSc thesis describing methods to generate large annotated datasets (hand segmentation) and a benchmark pipeline for local feature evaluation and SfM tasks.

Favicon imagedspace.library.uvic.ca

HandSeg: An Automatically Labeled Dataset for Hand Segmentation from Depth Images

January 1, 2019

Paper describing an automatically labeled hand segmentation dataset and associated methods for training deep segmenters.

Favicon imageabojja9.github.io

Patent filings (Microsoft) — e.g., "Interruption detection during automated workflow"

Inventor/co‑inventor on patent applications assigned to Microsoft Technology Licensing related to workflow interruption detection and related systems.

Favicon imagepaperdigest.org

Hobbies

Technical blogging and writing about computer vision and ML (Medium/personal blog). sriraghu

Open‑source development and sharing mobile ML demos (GitHub projects). github

Community teaching/mentoring (Kaggle, Udacity, BrainStation, StackOverflow contributions). kaggle+1