Skip to content
View AndrewHeller17's full-sized avatar

Block or report AndrewHeller17

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.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

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

Report abuse
AndrewHeller17/README.md

Andrew Heller

CS @ Dartmouth College, graduating June 2027. Currently studying abroad at the Aquincum Institute of Technology in Budapest.

I work at the intersection of AI, cybersecurity, and software engineering. Most of my recent work involves empirical research on ML models — evaluating reasoning performance, measuring bias, and trying to understand how these systems actually behave in practice.

andysh17@gmail.com - LinkedIn


Projects

LLM Emotional Framing Study — Evaluated how emotional framing affects LLM reasoning using McNemar/Cochran tests across MMLU and GPQA benchmarks. (Python, PyTorch)

Gender Bias in AI Image Generation — Measured gender bias in generative image models using chi-square and proportion tests. (Python)

Tiny Search Engine (private — available upon request) — A search engine in C with a crawler, indexer, and query engine. (C)

AI vs. Real Music Classifier (in progress) — Deep learning classifier to distinguish AI-generated music from real recordings. (Python, PyTorch)


Skills

Languages: Python, C, C++, Java, R, Swift

ML/AI: PyTorch, deep learning, statistical analysis, LLM evaluation


Experience

Teaching Assistant — Dartmouth CS50: Software Design & Implementation

Corporate Analytics Intern — Intermedia Cloud Communications

Technical Intern — PredictiveHR


Currently looking for SWE and cybersecurity internship for Summer 2026.

Pinned Loading

  1. Effect-of-Emotional-Framing-on-LLM-Performance Effect-of-Emotional-Framing-on-LLM-Performance Public

    Evaluated the impact of emotional prompt framing on LLM reasoning accuracy across industry benchmarks (MMLU, GPQA) using controlled experimental conditions.

    Jupyter Notebook

  2. Gender-Bias-from-Memories-in-ChatGPT-Image-Generation Gender-Bias-from-Memories-in-ChatGPT-Image-Generation Public

    Designed a three-stage pipeline to quantify demographic bias in AI image outputs by conditioning prompts on synthetic user memories referencing specific college majors, enabling controlled comparis…

    Jupyter Notebook