Open to opportunities

Magic Math

@magicmath

Message

Computer science and machine learning student focused on ML systems.

United States

Message

What I'm looking for

I seek roles where I can build and productionize ML systems, improve retrieval and evaluation pipelines, work in collaborative research-to-production teams, and grow in ML engineering and infrastructure.

I am a computer science and mathematics student with hands-on experience building ML systems, retrieval pipelines, and scalable web apps. I combine rigorous academic training with practical internships to deliver measurable improvements in performance and reliability.

At NIST I engineered a retrieval-augmented generation pipeline using LangChain and ChromaDB that improved retrieval accuracy by 12% and reduced latency by 30%, automated ingestion of 10k+ multimodal documents, and deployed REST APIs for real-time researcher dashboards.

I have contributed to ML evaluation and research projects—benchmarking GraphRAG and variants, implementing automated evaluation pipelines, and experimenting with multimodal time-series foundation models—while improving ETL throughput and reproducibility for team experiments.

Individually and in teams I build production-capable projects (NextJS, FastAPI, Supabase) and agentic prototypes that scale to 10k+ records; I prioritize automated testing, reproducible experiments, and delivering clear, deployable results.

Experience

Work history, roles, and key accomplishments

AI/ML Intern

National Institute of Standards and Technology

May 2025 - Aug 2025 (3 months)

Engineered a RAG pipeline (LangChain, ChromaDB) that improved retrieval accuracy 12% and reduced latency 30% on a 4k+ article knowledge base; automated ingestion of 10k+ multimodal documents and deployed REST APIs to stream wildfire/weather data for real-time researcher dashboards.

Lang Chain ChromaDB Docling REST APIs Tableau Python

Machine Learning Evaluation Intern

ML@UVA

Nov 2024 - May 2025 (6 months)

Collaborated with a 6-person team and client LMI to benchmark GraphRAG using BLEU, LLM-as-a-judge, cost, and latency; results were adopted for pilot trials and used to compare RAPTOR, CRAG, Self-RAG, and LightRAG.

BLEU LLM As A Judge Graph RAG CRAG Python

Machine Learning Researcher

Humanity Unleashed

Nov 2024 - May 2025 (6 months)

Researched multimodal time-series foundation models, experimenting with transformer and VQ-VAE variants via grid search; built ETL pipelines for 50GB+ datasets improving preprocessing throughput 40% and created reproducible testing pipelines with automated logging and visualization.

Transformers VQ VAE ETL data preprocessing Grid Search Experiment Logging Python

Education

Degrees, certifications, and relevant coursework

University of Virginia, School of Engineering and Applied Science

Bachelor of Science, Computer Science and Mathematics

2024 -

Grade: 4.0/4.0

Activities and societies: MARS; ML@UVA; Project Code; Relevant coursework: Data Structures and Algorithms 1/2, Discrete Math and Theory 1, Ordinary Differential Equations, Probability, Computer Systems and Organization 1, Software Development Essentials

Pursuing a Bachelor of Science in Computer Science and Mathematics with a 4.0 GPA and coursework in algorithms, discrete math, probability, systems, and software development; active in MARS, ML@UVA, and Project Code.

Thomas Jefferson High School for Science and Technology

High School Diploma, High School - Science and Technology

2020 - 2024

Grade: 4.394/4.0, SAT 1540

Activities and societies: Relevant coursework: AP Calculus BC, Multivariable Calculus, Linear Algebra, Artificial Intelligence, Machine Learning

Completed high school with advanced STEM coursework and strong academic performance (GPA 4.394/4.0, SAT 1540), focusing on calculus, linear algebra, AI, and machine learning.