M.S.E. in Computer and Information Science

• Coursework: Internet and Web Systems, Database and Information Systems, Natural Language Processing, Machine Learning

B.S. in Computer Science; Second Major in Mathematical Science

• CS GPA: 3.98 / 4.00
• Overall GPA: 3.84 / 4.00
• Focus: full-stack development, systems, security, mobile apps, and OS labs

• Built and evaluated retrieval-augmented generation workflows for DARPA SciFy and OpenScholar using score-based filtering, contrastive model work, and domain data preparation.
• Turned research goals into reusable scripts, datasets, and documentation so model and retriever changes could be tested consistently across the lab.

• Built an LLM-powered data portal with Danswer AI to search datasets from AWS, Datarade, and Snowflake across ingestion, indexing, and user-facing discovery workflows.
• Wrote Python crawlers and cleaners for 5,000+ datasets; resolved metadata issues that affected search relevance and downstream data quality.
• Automated AWS search infrastructure with CDK and deployment scripts for reproducible development, testing, and release support.

• Improved model robustness in glare and shadow conditions through targeted data augmentation, experiment tracking, and evaluation.
• Built a ROS accelerometer driver for sensor integration and applied self-attention distillation to improve convergence speed by 20%.

• Designed Python ETL modules and a unified schema for 10+ semiconductor clients, supporting analytics and reporting workflows.
• Reduced data processing time by 50% through modularization, stronger error handling, and reusable parsing logic.

• Built test automation for quantized neural networks on embedded systems, with MD5 checks, concurrent multi-device deployment, and structured client reports.

sirui.dev — Full-Stack Portfolio on AWS EC2

A typed Next.js portfolio shipped end-to-end: app, Nginx reverse proxy, PM2 process manager, Let's Encrypt HTTPS, and a GitHub Actions deploy pipeline running on Ubuntu EC2.

SafeCommerce-LA

Authenticated dashboards, maps, charts, filters, and CSV export over a PostgreSQL schema that joins LA crime data with California retail sales records. Graded 99/100.

Distributed Web Search Engine

End-to-end distributed search platform with a custom crawler, indexer, PageRank, replicated key-value store, multithreaded web server, and an RDD-style compute layer named Flame.

PennOS — A Unix-Like Educational OS

C implementation of Unix-like OS features: shell commands, process scheduling, signal handling, and a PennFAT filesystem with full error-path testing.

MasterCard Onboarding Automation

Full-stack onboarding tool with structured questionnaires, input validation, standardized submissions, and backend APIs for client intake.

Kinexus — WUSTL Medical Research Application

Fixed Apple Watch sync issues, refactored REST APIs, optimized AWS RDS queries, and wrote developer documentation that reduced onboarding time for new contributors.