Karthik Mittal

15-210 Teaching Assistant @ Carnegie Mellon University School of Computer Science

About

I'm a masters student in machine learning at CMU. Currently, I'm developing an end-to-end system to automate course creation using LLMs under the TEEL Lab, and I plan on continuing my research in multimodal summarization algorithms under the Safe AI Lab. I have received certifications in AWS and ML, authored bioinformatics papers under professors, and developed proficiency in several coding languages, including Python and R. In my spare time, I teach children competitive programming and math at local nonprofits in the Bay Area.

Country

United States

City

San Francisco Bay Area

Industry

Computer Software

Skill

Amazon Web Services (AWS), Python (Programming Language), Machine Learning, Computational Biology, Deep Learning, C (Programming Language), Natural Language Processing (NLP), Genomics, Microsoft Azure, Data Analysis, TensorFlow, Financial Analysis, Consulting, Neural Networks, Data Cleaning, Git, Robotics, Acoustic Guitar, Black Belt, Piano Playing

Experience

15-210 Teaching Assistant

Carnegie Mellon University School of Computer Science

2024-8 - Present · 2 yrs 2 mos

Teaching Assistant for Fall 2024, Fall 2025, Spring 2026. Head Teaching Assistant for Spring 2025.

Undergraduate Research Assistant - TEEL Lab

Carnegie Mellon University School of Computer Science

2023-5 - Present · 3 yrs 5 mos

Pittsburgh, Pennsylvania, United States

Developed Python autograder and evaluation rubrics for assignments in Python course (2,000 students). Transitioned from virtual machine submissions to an in-browser sandbox, increasing student satisfaction by 23%. Added JSON parsing and integrated exception handling in Java autograding infrastructure to process incomplete assignment submissions. Deployed these modifications in all active courses. Currently working on utilizing GPT4 to develop precise MCQs in curriculum.

Software Engineer Intern

Stripe

2025-5 - 2025-8 · 4 mos

San Francisco, California, United States

Refactored legacy components and integrated user state protobufs to support pinned sandboxes on Stripe Dashboard via React and GraphQL, improving sandbox retrieval time by 22%. Validated through A/B testing.

Software Engineer Intern

Coinbase

2024-5 - 2024-8 · 4 mos

San Francisco, California, United States

Added concurrent fraud checks for debit card transactions using Go, preventing $15K / day in malicious activity. Helped risk analysts identify root causes of fraud on Snowflake by adding granular transaction decline reasons. Designed PostgreSQL table to queue pending transactions for manual inspection, addressing $XM of fraud refund attacks. Altered webhooks code and built async job to process refunds, reducing latency by 12%.

Back End Developer

LatchBio

2022-6 - 2022-8 · 3 mos

Built several end-to-end bioinformatics workflows to simplify current digital computational biology tasks. These include a sequence alignment to UPGMA tree workflow and secondary structure prediction workflow using RNAfold.

Machine Learning Intern

Ellumigen

2021-6 - 2021-8 · 3 mos

Ocean’s advanced AI-driven platform extracts high dimensional transcriptomic features from RNA-seq data and combines it with clinical metadata to learn models and identify candidate multidimensional biomarkers. Optimized runtime for binary classification model through feature selection methods on RNA-seq data. Also worked with genetics PhD, finding recursive feature elimination and Fisher score to produce optimal accuracy.

Software Engineer

Mpathy Software

2021-4 - 2021-6 · 3 mos

Union City, California, United States

Mpathy EHR provides support for clinicians through an optimized electronic health record system having natural language processing and other modern features of artificial intelligence. This new EHR will replace traditional systems, minimizing the need to master complex user interfaces. Furthermore, this EHR will provide heightened recommendations according to inputted patient data. Developed autocompleter, speech recognition, and NLP components of the record system. Also worked with the CEO to ideate clinician-friendly additions, including color scheme and layout using PHP.