Looking for a job

Long Nguyen

@longnguyen8

Message

I build high-performance ML inference, diffusion, and RAG systems across C++, CUDA, Python, and JAX.

United States

Message

What I'm looking for

I'm looking to build high-performance ML, inference, diffusion, or RAG systems where I can turn research ideas into measurable improvements in speed, scalability, and workflow automation.

I'm advancing ML inference and diffusion research at UMass Amherst with Prof. Justin Domke and Prof. Yao Li, building faster methods for probabilistic models and score estimation.

I co-first-author a paper on compiler-style graph transformations that automatically apply JAX vmap, delivering an 11.5x net speedup in posterior estimation. I also built a cross-platform Pangolin PPL inference backend that integrates JAGS for Bayesian inference from plain Python.

At an AI × healthcare startup, I engineered RAG agents and document-intelligence pipelines that cut manual processing time by up to 75% and query latency by 50%. I built stateful workflow orchestration with LangChain and PostgreSQL to support dynamic branching, rollback, and resumable automation.

I also build GPU-accelerated systems, including a CUDA score-estimation pipeline that reached a 9x speedup over a CPU baseline on an A100. My projects span full-stack RAG applications, graph-based retrieval, and learning platforms built with React, FastAPI, C++, Java, Neo4j, and Docker.

Experience

Work history, roles, and key accomplishments

Current

Research Assistant - ML Systems / Inference

Current

UMass Amherst

Sep 2025 - Present (1 year)

Co-first-authoring a paper on compiler-style graph transformations for vectorization, yielding an 11.5x speedup in posterior estimation. Designed a hash-based algorithm and custom heap data structure to optimize large computation graphs, and built a cross-platform inference backend integrating JAGS for the Pangolin PPL.

Jax Python C JAGS Bayesian Inference (PyMC)Optimization Vectorization Data Structures

Research Assistant - Generative Model / Diffusion

UMass Amherst

May 2026 - Aug 2026 (3 months)

Researching score matching methods for diffusion models and implementing Monte Carlo density-gradient estimators in C++/Python for SDE systems. Optimized simulation throughput via SIMD-vectorized and fast RNG batch computation, cutting sampling latency by 20%.

C++Python Matlab SIMD Xoshiro128++Monte Carlo Diffusion Models SDE PDEs

Software Engineer Intern

Stealth Startup in AI × Healthcare

Jun 2025 - Sep 2025 (3 months)

Engineered two RAG agents for workflow automation using asyncio and vLLM, cutting manual processing time by 60% and query latency by 50%. Built a document-intelligence pipeline with vector similarity search and OCR, and designed a stateful workflow-orchestration service using LangChain and PostgreSQL.

Python AsyncIO VLLM Lang Chain PostgreSQL OCR OpenCV Vector Search RAG

Education

Degrees, certifications, and relevant coursework

University of Massachusetts, Amherst

Bachelor of Science, Computer Science and Math

Grade: 3.95

Activities and societies: Competitive Programming Tournament - First Prize (Sep. 2023); ICPC - Third Prize National, Asia-Pacific Qualifier, Regional Qualifier (Nov. 2023, Nov. 2024); Coursework: Reinforcement Learning, Machine Learning, Advanced Linear Algebra, Statistics and Probability

Pursuing a BS in Computer Science and Math with a GPA of 3.95, expected graduation in December 2027.