Available for work
Building AI systems and back-end services that turn complicated problems into software people actually want to use.
Software engineering student at Rochester Institute of Technology, specializing in applied artificial intelligence and back-end development. President of the RIT Artificial Intelligence Club, a 500-member student organization.
About
I'm studying Software Engineering with minors in Artificial Intelligence and Finance. That looks like hedging until you notice what both fields are actually about: making decisions under uncertainty, at speed, from incomplete data. FinRobot came directly out of that overlap: an agent doing the research legwork that analysts do by hand. The pairing wasn't a backup plan; it was the whole idea.
Most of what I find interesting isn't the model call; it's everything around it. Market APIs disagree with each other. Language models are confidently wrong. Study material arrives as a messy PDF. The demo is easy; making it hold up on the second run is the actual work, and it's where I spend most of my time.
As President of the RIT Artificial Intelligence Club, I run a 500-member organization, and the part I didn't expect to be doing is sales. I cold-pitched technology companies until one said yes, landing the club's first corporate sponsorship, and now coordinate with them to bring in engineers to give technical talks. Turns out explaining why your work matters is a separate skill from doing it.
Selected Work
Adaptive learning · In active development
NotebookLM could answer questions about my notes but had no memory of me, and no sense of whether I was improving. CramAI generates study material from your own documents and schedules it with FSRS-4.5, so reviews land as recall decays rather than on a fixed loop. Finding a prompt-cache ordering bug that was silently costing 10× on input cut up to 70% off a round.
Built for Discrete Mathematics for Computing, in use this semester
Read the case studyMulti-agent pipeline
Type a stock ticker, get a formatted research report. Three MCP servers running three different model providers: Perplexity for retrieval, Claude for formatting, Matplotlib for charts because charts shouldn't be a model's job, all coordinated by a Gemini orchestrator. It now refuses to draw a chart it can't get real prices for, which is the fix I'd defend hardest. Built with the RIT AI Club.
Presented to 200–300 attendees at ImagineRIT 2026
Read the case studySelf-directed ML curriculum
A machine learning curriculum I set for myself, running every stage on the same 5,000 movies instead of a new toy dataset per topic. Data wrangling through neural networks and recommenders, including the stage where I rebuilt a working classifier as a neural network and it came out worse.
Design & build
The site you're reading: no framework, no build step, no dependencies, 74 KB on first load. Rebuilt three times to find a voice that reads as engineer rather than enthusiast, and it taught me more about progressive enhancement than any tutorial has.
Stack
Experience
Fall 2026 to present
RIT Artificial Intelligence Club
Lead a 500-member student organization overseeing club operations, event programming, and workshops that build a campus community around applied AI research and development.
2026
Professional Certificate
Professional certification covering generative AI engineering, from model fundamentals through building and deploying LLM-backed applications.
Aug 2025 to May 2030
Rochester Institute of Technology
Pursuing a Bachelor of Science in Software Engineering, with anticipated minors in Artificial Intelligence and Finance.
Jul 2023 to present
Wegmans
Assist 10–20 customers per shift with product inquiries and recommendations while maintaining displays for freshness and presentation. Resolve customer issues by diagnosing needs and offering substitutions, building repeat business and trust.