Sean Patten

Computer Science | AI Research

1-934-500-4208
spatten2@fordham.edu

Resume (PDF) (DOCX)
LinkedIn
GitHub
Games
Stories
Call
Chat
Search
Realtime
Maps
Unix V7
Linux
Links
Ideas
Drafts
Beta

Experience

Member, Laboratory for Informatics and Data Mining (Dr. D. F. Hsu), Fordham (2024 – Present)

Fordham CIS research lab for machine learning, data mining, and information fusion, centered on combinatorial fusion analysis.

• Research Team Lead (2026 – Jun 2026): 40-model quantum-classical combinatorial fusion pipeline, with Dr. Samuel Yen-Chi Chen (first author of the original QLSTM paper).

AI Engineer, Aspire Aviation (Jan 2025 – Present)

• Predictive models for aircraft part failure; LLM document generation.
• Constraint-based pilot scheduler in live use.

Software Developer (Contract), SSRC (Oct 2024 – Jan 2025)

• Data pipeline on LLM adoption trends on Hugging Face.
• Under direction of Tim O'Reilly and Ilan Strauss.

AI Research Intern, IBM (May 2023 – May 2024)

• Led ensemble system for sentiment analysis using combinatorial fusion.
• SOTA 97.072% on IMDB.
• Peer-reviewed publication (PICom 2025), basis of Master's thesis.
• Collaborators: Dr. Pin-Yu Chen, Dr. D. F. Hsu, Dr. C. Schweikert.

Publications

Xu, J., Wu, Y., Chen, S. Y.-C., Patten, S., Schweikert, C., Hsu, D.F. (2026). Diversity-Aware Ensemble of Quantum-Classical Models via Combinatorial Fusion. QCE 2026, Toronto (accepted).

Patten, S., Chen, P.-Y., Schweikert, C., Hsu, D.F. (2025). Enhancing Sentiment Classification with Machine Learning and Combinatorial Fusion. PICom 2025. [arXiv] [Code]

Erdos Number: 2 (via Dr. D. F. Hsu)

Projects

a: agent fleet & automation platform

Compiled-C CLI, systemd job hub, git-synced state across a device fleet, orchestrating AI coding agents with review gates. Serves this site's dynamic pages through a Cloudflare tunnel.

SuperCowsay: cowsay rewritten in assembly, 1.3 µs above the Linux exec floor

629-byte hand-written ELF (assembler emits it directly, no linker): 1.3 µs above a bare exit(0), 145x faster than the Perl original, 2 syscalls. A second build is byte-identical to Perl across 344 differential-fuzz cases at 35.7x its speed.

e: terminal editor in C, faster than Torvalds's uemacs

0.37 ms startup, 2.8x faster than uemacs; opens a 5,600-line file in 1.2 ms (nvim: 223 ms).

This site

Single-file, 100/100 PageSpeed, 1991 HTML tag subset. Two-player pong with no game server (WebRTC peer-to-peer) and a real AI chat (Gemma 4 on my own machine, CPU only) served by a's C webserver.

Custom automated trading system

In production with real capital; proprietary and cannot be shown or described in detail.

Research Objective

Training, combining, and architecting ensemble strategies to improve performance, safety, and cost effectiveness.

Education

M.S. Computer Science, Fordham (Aug 2022 – May 2024)

• GPA 3.667/4.0
• Thesis: Enhancing Sentiment Classification with Machine Learning and Combinatorial Fusion
• Advisor: Dr. D. F. Hsu
• Coursework: AI, Deep Learning, NLP, ML, Data and Information Fusion

B.S. Computer Science & B.S. Business Administration, Fordham (Aug 2018 – May 2022)

Awards

• National Merit Scholarship Semifinalist
• Johns Hopkins Center for Talented Youth Student
• We the People 2018 NY State Winner & Nationals Finalist

Technical Skills

AI Agents: Claude Code, OpenAI Codex, Gemini CLI, Grok, Google Antigravity; used together in multi-agent workflows: worktree orchestration, review gates, MCP, browser automation, device fleets, LLM fine-tuning, RAG

Languages: Python, C, C++, PyTorch, Tensorflow, Java, Javascript, SQL, CUDA, Assembly

Platforms: Kubernetes, Qiskit, Huggingface, Git

Teaching

Club President, Fordham Digital Business Society (2021 – 2022)

Taught Python to beginners, increased attendance over 1,000%.

Quantum Computing

Quantum Computing Researcher, Fujitsu Quantum Simulator Challenge (Jan 2026 – Present)

• First author of the Asia University team's accepted proposal, Q-Compress: Variational Entanglement Profiling for Extreme LLM Compression.
• Running LLM weight-compression experiments on Fujitsu's 40-qubit tensor network simulator, a 1,024-node FX700 (A64FX) supercomputer cluster.