SOFTWARE ENGINEERING & INTELLIGENT SYSTEMS
Software that thinks.
Systems that hold up.
FROM THE FIRST TENSOR TO THE FINISHED PRODUCT.
01 / SELECTED SYSTEMS
Four builds. From the interface people use to the infrastructure that makes it work.
Personal AI assistant
Email, calendar, coursework, and subscriptions—connected through a conversational assistant with voice, approvals, and durable workflows.
Deep learning systems
A deep learning engine built from first principles, with its own gradients, trainable layers, GPT model, and native execution experiments.
Portfolio research platform
Portfolio planning and historical research with validated prices, exact accounting, immutable inputs, and independent calculation checks.
RELEASE CONTROL / MODEL LIFECYCLE
ML deployment infrastructure
A model registry and deployment control plane with immutable artifacts, weighted canaries, promotion, rollback, and runtime boundaries.
02 / INDEPENDENT RESEARCH
TOOLRET / RETRIEVAL STUDY
When does tool reranking earn its cost?
I studied selective reranking over a catalog of 37,292 tools, separating development, calibration, and confirmation before evaluating a learned router on untouched queries.
351 fewer calls on untouched confirmation queries.
Predeclared nDCG@10 tolerance: 0.01. Quality superiority was not established.
03 / BEHIND THE BUILDS
const michael = {
school: “Cornell”,
degree: “Computer Engineering”,
graduation: 2028,
interests: [“AI”, “systems”]
};
I’m Michael, a Computer Engineering undergraduate at Cornell University. I build across the stack—from the math behind a gradient to the software around a reliable service.
I like problems that don’t stop at the demo. How does the system recover? Can someone reproduce the experiment? What does the user need to understand? Those questions shape my work.
College of Engineering
B.S. Computer Engineering · Expected May 2028Python · TypeScript · C/C++ · Go · Java · SQL
PyTorch · NumPy · Transformers · Retrieval · Autograd
Docker · PostgreSQL · MongoDB · Temporal · AWS · Terraform
04 / EXPERIENCE & COMMUNITY
Collaborate with staff and student researchers, train students in algorithms and AI/ML, and coordinate programs for 100+ Cornell students. Automate finance and reporting workflows while handling program data securely.
Built a public-data mpox tracker covering 80+ countries with JavaScript and Google Charts. Studied algorithms, APIs, and debugging in a four-week program with Google engineers.
2026Designed a disjoint retrieval study, froze evaluation decisions before confirmation, and measured the cost–quality tradeoff of conditional reranking.
Break Through Tech AI/ML Fellow · Bloomberg Tech Fellow · Cornell Engineering Student Ambassador · Cornell Orientation Leader
05 / START A CONVERSATION
Software and AI/ML opportunities, technical collaboration, or a good conversation about systems.
