Hi, I’m Meron 👩🏽💻
Undergraduate at Smith College studying computer science and applied mathematics, graduating May 2028. My work sits at the intersection of machine learning research, interpretability, and building systems that are rigorous and useful in practice.
Research interests
I'm most drawn to mechanistic interpretability — understanding what happens inside language models by tracing attention patterns, internal representations, and feature circuits. More broadly, I care about multi-agent safety and AI governance: what does it mean to understand a model well enough to trust it, and how do we audit systems that behave in ways we didn't explicitly design?
I recently replicated key results from In-context Learning and Induction Heads (Olsson et al.) using TransformerLens on GPT-2 Small.
Current work
- Undergraduate researcher in Smith’s High Performance Computing Lab
- Studying optimization framework behavior and neural network training dynamics
- Extending this work toward loss landscape interpretability
Background
AI Engineering Intern
Built a Random Forest classifier on 5,000+ workflow events (F1 = 0.835) and a real-time Streamlit dashboard that reduced manual review time by 50%.
Conway Innovation and Entrepreneurship Center
Built an offline-first Next.js application for tracking community program delivery in low-connectivity environments.
Machine Learning Researcher
Built an end-to-end clinical audio ML pipeline across 864 recordings from 108 patients, with early interpretability work in a high-stakes setting.