Education

---

• GPA: 3.96/4.00
• Dean’s Scholarship, All Semesters
• University of Rochester Bond Scholar, 2026
• Coursework: Advanced Algorithmic Design, Introduction to AI/ML, Formal Computational Models and Limitations, Data Structures and Algorithms, Linear Algebra, Differential Equations

Work Experience

---

• Developed SDMComms, an internal web portal using Angular, JavaScript and ServiceNow that serves over 90 developers across service delivery management teams, consolidating monthly newsletters, announcements, and platform updates in one place
• Integrated email functionality within the portal that allows administrators to send communications directly to end-users without leaving the ServiceNow platform, reducing managerial communications work by ~50%
• Scoped project from ideation to execution by conducting stakeholder interviews across 5 platform teams to define feature requirements, then translated requirements into a prioritized roadmap and agile sprint backlog executed over 10 weeks

• Invented and published the Inverse Turing Benchmark, a Hugging Face LLM Benchmark that evaluates model submissions over 557 unique dialogue pairs on ability to recognize if a given conversation is human/AI or AI/AI
• First author of InverseTuringBench, a paper outlining the Inverse Turing Benchmark and related findings regarding semantic and statistic AI-text detection amongst 29 detection methods (Under review, EMNLP 2026)
• Contributed to development of HAL (Human Aligning LLM), a framework for LLM alignment with emphasis on inducing human-likeness traits, increasing win-rate on a human-likeness Turing Test from 34% to 61%

Projects

---

• Designed and developed a dice-rolling RPG video game with Godot engine and GDScript (Python) to implement 24 unique turn-based combat encounters, 3 playable characters, and 14 customizable dice
• Used Godot’s built-in scene architecture and signal system to handle transitions between 7 game scenes, managing tooltips and global state using object-oriented principles
• Optimized no-SQL database and data retrieval system by implementing a modular management layer, reducing load time by ~66% and improving player experience
• Established and adjusted core game mechanics with a small-scale playtesting group of around 10 early users in order to ensure balanced and enjoyable gameplay, as well as dynamic difficulty scaling

• Architected and implemented a full stack web application that averages ~100 weekly visits for the University of Rochester BAJA Society of Automotive Engineers club, using a React + React Router + MongoDB stack
• Created 6 idempotent RESTful endpoints using JavaScript and Netlify serverless functions to ensure dynamically scaled performance based on current user loads
• Built Auth0 protected schema-based Content Management System, allowing non-coding club leaders to update content without code changes

• Engineered CatGPT, an application that uses an open-source transformer model (Xenova MiniLM, via Hugging Face) to extract 768-dimensional vector embeddings from user-submitted text input
• Implemented cosine similarity comparisons between generated embeddings and 10 pre-computed embeddings for food-related anchor words to quantify semantic relevance on a continuous numeric scale
• Devised an exponential scaling function to convert similarity scores into a variable number of output "meows," translating embedding-based similarity into an interpretable result

Skills

---

• Programming Languages: Java, C, Python, JavaScript, GDScript, HTML, CSS
• Frameworks:React, Angular, Gradio, scikit-learn, Java Swing, Java AWT, JavaFX, LibGDX
• Technical Skills:MongoDB, Git, Linux/Unix, Valgrind, Object-Oriented Programming, LLM Alignment, ServiceNow