I'm a junior at Amherst College studying Computer Science & Mathematics. I'm passonate about coding, computational research, and backend systems. I'm open to internship oppertunies — please feel free to reach out at rklopfenstein27@amherst.edu

Feel free to check out some of my work below! ⬇️

Currently building Stealth AI Startup.

• verieql-4-text2sql (Python, SQL, SQLite, PostgreSQL, MySQL, veriEQL, Regex, Amherst HPC) -- A two-stage Text‑to‑SQL system that first generates schema-aware SQL using LLM-powered components (TASL and TALOG), then applies formal verification with the veriEQL framework to ensure semantic equivalence across different SQL dialects. Just accepted to MATH-AI, a workshop at NeurlIPS!
  
  
• "Drop AI" (Python, Flask, REST APIs, Selenium, BeautifulSoup, OpenAI API) -- A full-stack platform that transforms raw content into actionable intelligence using Gemini 2.5 Flash. The Flask backend powers email classification, PDF extraction, and multi-function AI pipelines, while the Next.js frontend provides a modern, responsive UI. Key features include automated email replies, PDF summarization, and intelligent routing of content, delivering results in seconds for professionals who need streamlined workflows.
  
  
• "Tagging The Switch" (Python, PyTorch, Hugging Face Transformers, XLM-R, RemBERT, Pandas, Matplotlib) -- A multilingual NLP project exploring POS and language‑ID tagging in English‑Spanish code‑switched tweets. It compares single‑task, sequential, and multi‑task training strategies using XLM‑RoBERTa/RemBERT, including dynamic uncertainty‑based loss weighting.
  
  
• Image Edge Detector (Java, Swing, AWT, Sobel Filters, BMP Handling) -- A Java desktop app with a GUI that applies Sobel‑based edge detection to BMP images, offering features like grayscale conversion, random edge coloring, and a unique "Stars" edge outline mode.