Sean Kim

Rutgers student passionate about building accurate NLP models for resource-constrained environments with experience in full-stack web development, DevOps, and MLOps.

education

Rutgers University-New Brunswick

B.A. Computer Science

Sep 2022 - May 2026

• GPA: 3.88
• Dean's List: 2022-2024
• Coursework: DSA, Deep Learning, Databases, Data Management, Computer Architecture

experience

Full Stack Developer

ACL Agency LLC

Jun 2024 - Aug 2024

• Rebuilt company website with SvelteKit, adding SEO-friendly Korean language support with SSR which boosted site traffic by 30% and enhanced market presence in the local community.
• Optimized an internal Go service with goroutines, achieving a 42% increase in insurance claims processing speed on over 500 concurrent OCR and file compression tasks.
• Deployed service on AWS EC2 with Nginx for SSL termination and whitelisting using Docker Compose.

projects

grammaCy

Python, spaCy, Flask, Nginx, Docker, AWS EC2, SvelteKit, Firebase

• Led a team of 3 to develop a grammar-checking library with spaCy featuring a CoNLL-U corpus augmentor that improved processing speeds on OntoNotes 5.0 by over 10x with parallel and concurrent processing.
• Built a prepackaged English model by training a dependency parser (93% LAS) and POS tagger (98% accuracy) on an augmented GUM corpus, checking over 20 grammar errors with 83% accuracy.
• Constructed a Flask API endpoint running on Gunicorn to serve over 600 concurrent English grammar and spell-checking requests, averaging 25ms prediction times.
• Deployed the endpoint behind an Nginx reverse proxy for SSL termination and rate limiting on AWS EC2 using Docker Compose.
• Created a SvelteKit site for library documentation and API guides, deployed on Firebase Hosting.

ref:note

Go, TypeScript, Node, Next, OAuth2, Docker, GCP, Vercel

• Led a team of 2 to develop a collaborative note editor with Next, deployed on Vercel.
• Implemented real-time collaboration using Yjs on a Node WebSockets server running on Google Cloud Run, handling 250 concurrent note edits with 100% consistency between clients.
• Built a Go backend for database operations on Google Cloud Run, reducing cold start times to 100ms.
• Secured API endpoints using OAuth2 and JWT, enabling 1000 users to safely access collaborative features.
• Leveraged Firestore to store user and note data, implementing granular access control for each note.

skills