Tommy Nguyen

Machine Learning Engineer & Independent Researcher

Research Interests: Machine learning systems, distributed training, coding agent harnesses and multi-agent coordination, evaluation benchmarks, and AI economics.

Education

Drexel University

B.S. in Computer Science, Pennoni Honors College (5-Year Co-op Program)

Concentrations: Artificial Intelligence, Computing Systems

GPA: 3.88 / 4.00 · Magna Cum Laude · Dean's List

Selected Coursework: Machine Learning, Artificial Intelligence, Theory of Computation, Operating Systems, Concurrent Programming, Systems Architecture, Software Architecture I, Database Systems, Programming Language Concepts, Data Structures, Systems Programming, Game AI Development, Linear Algebra, Multivariate Calculus, Probability & Statistics, Discrete Mathematics, Senior Project I–III (CI 491–493), NER Research Project (CS I499)

Walk with Code — Scratch to Scale: Real-World HPC Training

Distributed Training Course (Instructor: Zachary Mueller, Maven)

Implemented distributed training algorithms from scratch across multi-GPU and multi-node HPC clusters, covering parallelism strategies, memory and communication overhead, and performance profiling and debugging.

Illinois Math and Science Academy (IMSA)

High School Diploma · GPA: 3.75 / 4.00

Research, Manuscripts & Writing

Working Manuscripts & Academic Research

  • Tommy Nguyen, Marcos Didonet Del Fabro, and Hegler Tissot. "FOREST: Phonetic Trie Ensembles with Logistic Regression Filtering for Low-Resource Approximate Named Entity Recognition." Working Manuscript, Drexel University (CI 491–493 / CS I499), 2022. [PDF]
    Code and datasets in nwyin/forest.

Abstracts & Presentations

  • Tommy Nguyen. "Augmented Reality Visualizations for Software-Defined Radios." Drexel University 2019 STAR Scholars Summer Showcase Abstract Booklet, Office of Undergraduate Research, Pennoni Honors College, Philadelphia, PA, August 29, 2019, p. 123.
    Faculty Mentor: Dr. Kapil R. Dandekar (Electrical & Computer Engineering, Drexel Wireless Systems Laboratory).

Selected Essays & Technical Writing

Experience

Independent Research & Open-Source Engineering

Independent ML Systems Researcher & Engineer

  • Designing and evaluating coding agent harnesses (tau), multi-agent orchestrators (hive), and empirical micro-benchmarks (edit-bench, meta-rlxar-reproduction) to study edit strategies, post-training rubrics, and parallel agent coordination.
  • Completed distributed training coursework (Scratch to Scale) and scaled a 124M GPT-2 codebase to a 7B Llama-2 run using PyTorch FSDP (modded-llama2).
  • Shipping open-source Rust and Python developer tools (irradiate, pycg-rs, pycfg-rs, billwatch, slop-guard) alongside interactive AI economics and scientific data explorers on nwyin.com, while actively collaborating and exchanging notes with frontier AI engineers, researchers, founders, and investors.

Nuuly (Urban Outfitters)

Senior Machine Learning Engineer

  • Led development of transformer-based user and item embedding models to power personalization and search.
  • Architected and developed microservices to scale e-commerce logistics processing, increasing order processing throughput and reducing warehouse labor costs.
  • Sped up CI/CD, set up experiment tracking infrastructure, and hosted weekly journal clubs to discuss novel work and libraries.

U.S. Census Bureau

Software Engineering Fellow, Civic Digital Fellowship

  • Developed web pages for U.S. Census Bureau content using React and AWS Amplify.
  • Prototyped a content management system (CMS) to be used for U.S. Census Bureau 2020 Census content.

Sharing Excess

Lead Software Engineer (Co-op)

  • Developed features for Google Firebase and a React web application, reducing the time needed to coordinate food rescue logistics and helping rescue over 200,000 lbs of food.
  • Onboarded pilot users onto the MVP of the application for testing and feedback.
  • Communicated business requirements and coordinated development efforts across six junior software engineers, business development, and end-clients.

Rise First

Web Developer Fellow

  • Developed and migrated content from a pre-built Squarespace site to a private repository using Node.js, Express, and Bootstrap.
  • Optimized pages and content for search engine rankings and accessibility, resulting in 2x organic search traffic.
  • Collaborated closely with marketing and content teams to align delivered features with end-user expectations.

Randstad (Vanguard Group Assignment)

IT Developer Intern (Co-op)

  • Cleaned and assisted with feature engineering for a dataset of over 700,000 entries to be analyzed by a machine learning model to guide business action.
  • Designed and developed automated processes using Blue Prism, Python, and AWS Lambda, saving over 50 hours of manual work per month.
  • Deployed and documented two automated processes into production using CI/CD tooling (Jira, Bamboo, Bitbucket, CodeCommit, CodeBuild).

Drexel Wireless Systems Laboratory (DWSL)

Undergraduate Researcher, STAR Scholars Program · Mentor: Dr. Kapil R. Dandekar

  • Built and deployed a Flask API on Docker to access real-time telemetry from Software-Defined Radios (SDRs).
  • Designed and developed an augmented reality mobile application using Unity and Vuforia to visualize real-time data transmission and radio availability from SDRs.

Selected Projects & Open Source

  • modded-llama2 Distributed Training · PyTorch FSDP

    Reproduction of @KellerJordan's modded-nanogpt trained on Hugging Face's FineWeb dataset, scaling from a 124M GPT-2 baseline to a 7B Llama-2 model using PyTorch FSDP and reaching <3.28 validation loss on 8×H100s in ~8 hours wall-clock time.

  • edit-bench Agent Evaluation · Micro-benchmark

    Mutation-based edit strategy benchmark evaluating string-replace vs. line-anchored hashline edits across Python, TypeScript, and Rust coding tasks.

  • meta-rlxar-reproduction Post-Training · RL Reproduction

    Open reproduction study (code, prompts, and logs) testing Meta's RL-XAR rubric-optimization method for prose quality.

  • hive Multi-Agent Systems · Python

    Single-process async orchestrator coordinating parallel coding agents against a SQLite task queue with isolated Git worktrees, shared notes, and an automated merge pipeline.

  • tau Coding Agent Harness

    Minimal coding agent harness built to explore tool use, edit formats, context management, and conversation loops.

  • irradiate Program Analysis · Rust

    Fast mutation testing for Python written in Rust — parses and mutates Python ASTs natively before executing the Python test runner.

  • pycg-rs & pycfg-rs Static Analysis · Rust

    Whole-program static call graph (pycg-rs) and per-function control-flow graph (pycfg-rs) generators for Python written in Rust.

  • billwatch CLI · SQLite + FTS5

    Cached, token-efficient CLI for querying U.S. state legislative data with agent-ergonomic structured outputs across four backends.

  • dw2md CLI Tool

    Crawls DeepWiki-generated documentation sites and consolidates pages into a single LLM-ready Markdown context file.

  • forest Approximate NER · Drexel Senior Project (CI 491–493 / CS I499)

    Rule-based approximate named entity recognition system combining phonetic trie ensembles (Double Metaphone, Metaphone, NYSIIS, Soundex) with a calibrated logistic regression filter, reaching 0.950 F1 on perturbed FDA drug names. Manuscript PDF.

Honors, Fellowships & Awards

  • Magna Cum Laude — Drexel University
    Dec 2022
  • Dean's List — Drexel University College of Computing & Informatics
    2018 – 2022
  • Civic Digital Fellowship — Coding it Forward · U.S. Census Bureau
    Jun – Aug 2021
  • SASE Scholarship (Lockheed Martin Sponsored) — Society of Asian Scientists and Engineers
    Sep 2020
  • Aspires Scholars — Drexel University
    Nov 2019
  • STAR Scholars Program — Drexel University Pennoni Honors College
    Jun – Aug 2019
  • Sigma Phi Epsilon Balanced Man Scholarship Finalist — Drexel University
    Oct 2018
  • National Kikkoman Scholarship Recipient — Study abroad in Japan with host family
    Jul – Aug 2017

Teaching, Mentoring & Leadership

Drexel University College of Computing & Informatics

Peer Mentor

  • Mentored 22 students in a small classroom setting on short-term academic and career goals.
  • Built online community programming during the COVID-19 pandemic via Discord, forums, and small-group meetings; hosted and recorded a career panel with four industry professionals.

Drexel University Housing and Residence Life

Resident Assistant

  • Planned and hosted community-development events for 38 freshmen in suite-style residence halls.
  • Coordinated with a 15-person staff to manage North and Caneris residence halls and ensure student safety and wellbeing.

IMSA Writing Center

Writing Tutor

  • Tutored 100+ students to improve clarity, grammar, and argumentation across literature, history, research papers, resumes, and college essays.
  • Provided 200+ hours of real-time editorial feedback in the Writing Center.

Society of Asian Scientists and Engineers (SASE)

Webmaster

Dragon's First

Alumni and Outreach Chair · Organization for First-Generation / Low-Income Students

Drexel Asian Students Association

Member

Additional Experience

  • Careers Assistant — Steinbright Career Development Center, Drexel University · Compiled graduate school checklists and supported staff logistics
    Apr 2019 – Jun 2019
  • Student Barista — Starbucks (Drexel Campus), Philadelphia, PA
    Jan 2019 – Apr 2019

Technical Skills

Languages
Python, Rust, C/C++, CUDA, SQL, TypeScript/JavaScript, Bash
ML / AI & Data
PyTorch (FSDP, distributed training), Transformer & embedding models, NumPy, Pandas, Polars, Apache Spark
Infrastructure & Tools
Google Cloud Platform (GCP), AWS (Lambda, Amplify), Docker, Kubernetes (k8s), Apache Airflow, Kafka, PostgreSQL, SQLite (FTS5), Git