I'm a natural general intelligence currently living in the Chicagoland area. A Computer Science and Mathematics graduate from the University of Illinois Urbana-Champaign, I specialize in building scalable software, optimizing algorithms, and exploring the depths of Mechanistic Interpretability to ensure we can flourish and exist peacefully with our machine peers.
Experience
Experience
Mechanistic Interpretability
Independent Study | June 2025 – Present- Exploring advanced AI safety through techniques like Activation Patching and Linear Probes.
- Conducting experiments on local models involving ablation and token prefill.
- Utilizing Sparse Autoencoders for model analysis.
Machine Learning Intern
HGS Digital | June 2024 – August 2024- Delivered AI solutions including graph-based RAG systems.
- Improved fraud detection models using One-Class SVM.
- Reduced false positives by 10% through model optimization.
KDE Contributor
Open Source | 2022 – Present- Collaborating with global developers on code development.
- Reviewing high-impact Open Source Linux applications.
- Contributing to the KDE ecosystem through various patches.
Apart from working on machine learning and software engineering projects, I also enjoy learning more about mathematics, including graph theory and complex analysis. Mathematics is particularly enjoyable due to its rich structure and the beauty of certainty of truth. I also enjoy learning about history, in particular early Christian religious practice following Jesus, as well as early Judaism. Outside academic interests, I engage in biking, watching sports and cooking.
Feel free to check out my notes, available at the link in the header.
Projects
Projects
Agent With Retrieval
OpenAI, Vector Databases, Structured Outputs, React- Built for retrieval across a large input corpus, through embeddings, clustering and linear operations in the latent space.
- Interaction through a React interface, utilizing structured outputs for multiple use cases, including storytelling and programming.
- Exploratory work on fine-tuning for optimizing subjective criteria.
Uniform Job Scheduler
Formal specification, Docker, Operational Research- Combinatorial optimizer (Google OR-Tools) used to approximate best schedule for scheduling bag-producing machines.
- Improved bag throughput by ~5% in simulation.
IO_URING Webserver & Logger (Serves this website)
C++20, io_uring, OpenSSL, Google Compute Engine, Linux Systems- Contains a high-performance HTTP server with TLS implementation, a MCP server for home automation integration, and an extensible backend for future protocols (FTP).
- Leverages the asynchronous io_uring Linux system call to handle multiple clients efficiently from a single-threaded event loop.
- Logs and analyzes traffic traversing a WiFi-Ethernet bridge colocated with the server.
- Utilizes Google Compute Engine for external communication and secure tunneling.
MCP Remote File Server
Agentic AI, MCP, Search & Indexing, GCP, Network Filesystems- Created a Model Context Protocol (MCP) server for accessing files on remote filesystems.
- Allows reading PDF, DOCX, and PPTX and includes features for reducing amount of context, including partial truncation of files.
- Maintains an efficient reverse index to speed up access due to speed constraints of network filesystem traversal.
Education
Education
University of Illinois Urbana-Champaign
MS Biomedical Imaging
Incoming Graduate Student | Fall 2026
University of Illinois Urbana-Champaign
BS Mathematics & Computer Science
High Distinction | GPA: 3.5 | 2021 – 2025
Relevant Coursework
- Systems Programming
- Database Systems
- Computer Architecture
- Program Verification
- Machine Learning