Karishma Mandyam

Senior Machine Learning Engineer at Meta, NLP Researcher

Greater Seattle Area

Summary

Karishma is a distinguished Machine Learning Engineer at Meta, specializing in Natural Language Processing (NLP) and AI for critical applications like content integrity. Her work involves developing advanced AI models to detect emerging violations and contributing to cutting-edge research in instruction following for Large Language Models (LLMs). medium+1

With a strong academic foundation from the University of Washington, where she earned both a BS and MS in Computer Science, Karishma has a deep background in NLP research. Her graduate work included a master's thesis on 'Semi-Supervised Learning For Verbal Autopsy' and significant contributions to the ARK group under Noah Smith. github+1

Karishma has extensive internship experience at leading tech companies, including two stints at Facebook focusing on NLP projects for Messenger Assistant and Newsfeed Integrity (with one project published at EMNLP), and two internships at Microsoft across Azure Security and Customer Experience Engineering. She also interned at Blue Origin, working on internal tooling. github

Beyond her technical roles, Karishma is a dedicated educator, having served as a Teaching Assistant for multiple Computer Science courses at the University of Washington. She also actively shares her experiences and insights through her personal blog on Medium, offering reflections on her career progression, personal growth, and various interests. github+1

Karishma is an advocate for women in STEM, having founded her local Girls Who Code chapter and participated in programs aimed at encouraging young women in technology. Her early accomplishments include winning a TiE Young Entrepreneurs competition for developing an innovative iPhone case to signal emergency services. github+1

Work

Education

Writing

Rubric-Based Benchmarking and Reinforcement Learning for Advancing LLM Instruction Following

December 1, 2025

Co-authored a research paper focusing on improving instruction following in Large Language Models (LLMs) through a novel post-training pipeline called RIFL, leveraging expert-curated rubrics for evaluation and training.

ai.meta.com

2022 Year in Review

December 31, 2022

A personal reflection on experiences, memories, and milestones from 2022.

medium.com

2021 Year in Review

December 31, 2021

A personal reflection on experiences, memories, and milestones from 2021, including working at Facebook as an ML engineer, getting vaccinated, family road trips, and solo adventures.

medium.com

2019 Year in Review

December 31, 2019

A comprehensive review of experiences, memories, and milestones from 2019.

medium.com

What Happened in the Last 525,600 Minutes?

January 1, 2019

A reflection on 2018, focusing on tackling challenges and building confidence.

medium.com

Three Insights from Ten Weeks at a Rocket Company

September 10, 2017

A computer scientist's perspective on her internship experience at Blue Origin.

medium.com

10 Reasons to Attend a Hackathon

A blog post encouraging participation in hackathons.

medium.com

Hobbies