← Back to profile
KO

Soojin Jung

Software Engineer

sojjung3@gmail.com · linkedin.com/in/soo-jin-jung · github.com/soojjung · medium.com/@sojjung3

Summary

M.S. student at New York University and frontend engineer with 3+ years of experience building production-grade web and mobile applications using React, TypeScript, TanStack Query, Zustand, and modern API-driven architectures. Experienced in AI-powered product development, RAG pipelines, real-time streaming interfaces, and production monitoring. Seeking software engineering internship opportunities in frontend, full-stack, AI platform, and MLOps frontend roles.

Education

M.S. in Computer Engineering (AI Robotics Track)

New York University

Sep 2026 — Present

  • Focus: AI systems, robotics, machine learning, and human-centered AI interfaces.

BSc. in Applied Mathematics and Statistics & Economics

Stony Brook, State University of New York — GPA 3.57 / 4.0

Graduated Dec 2018

  • Relevant Coursework: Intro to Computers, Finite Mathematical Structures, Linear Algebra, Probability Theory, Survey of Probability and Statistics, Econometrics, Operations Research.

Skills

AI / LLM Systems

RAG pipelines LangChain ChromaDB OpenAI API Fallback orchestration

Real-time & Data Interfaces

SSE Whisper Streaming UI Dashboard UI API-driven workflows

Reliability & Analytics

Sentry Production monitoring Incident response GA4 Mixpanel

Frontend

React Next.js TypeScript JavaScript TanStack Query Zustand Tailwind

Backend / Data

FastAPI Spring Boot REST APIs MySQL Supabase

Infrastructure / Tools

Docker AWS Vercel GitHub Actions Capacitor App Store Release

Work Experience (3 yrs)

Frontend Developer

BeHappy

Oct 2024 — May 2025 (8 mos) · Korea

  • Built an AI-powered mobile finance service that generates long-term financial roadmaps for newlyweds via interactive chatbot onboarding.
  • Integrated a low-latency (<500ms) streaming infrastructure using SSE (Server-Sent Events) and OpenAI Whisper to deliver real-time, interactive chatbot reports.
  • Engineered a 4-tier data persistence strategy using Recoil and local storage to manage 310+ dynamic fields, ensuring zero data loss during complex, multi-step user onboarding.
  • Eliminated data type mismatches in AI-model communication by implementing Zod superRefine for cross-field validation, significantly enhancing system-wide data integrity.
  • Spearheaded a data-driven product pivot using GA4 / Mixpanel analytics, revealing a 95% mobile access rate and high engagement in unexpected demographics.
JavaScript React Recoil Tailwind CSS Zod GitHub Actions

Frontend Developer

aiZENGlobal

Mar 2022 — Jun 2024 (2 yrs 4 mos) · Korea

  • Developed a loan brokerage platform that helps e-commerce sellers unlock financing by leveraging receivables from marketplaces like Coupang as collateral.
  • Architected and deployed a tri-portal ecosystem from the ground up, managing distinct interfaces for Customers, Internal Operations, and Financial Partners.
  • Engineered a configuration-based SearchTable pattern and a library of 99+ reusable UI components, reducing development time for 30+ administrative pages by 60%.
  • Implemented real-time data visualization dashboards for collateral verification, integrating backend scraping APIs to enable instant loan eligibility assessments for users.
  • Developed a JWT auto-renewal system with a Request Queue pattern and an 8-stage secondary authentication flow to secure mission-critical financial data.
JavaScript React Context API Styled-components Ant Design Axios

Selected Projects

AI Wellness App for Women's Health (dwee)

github.com/soojjung/dwee-wellness-tracker

Apr 2026 — Present

  • Building a women's wellness app for menstrual cycle tracking, next-cycle prediction, condition logging, and future AI-based exercise, nutrition, and mood-journaling recommendations.
  • Designed the MVP around cycle prediction and user feedback collection, with a roadmap to compare rule-based, statistical, and ML-based prediction models.
  • Established an AI-assisted engineering workflow using repository-level context, coding rules, custom commands, memory control, and sub-agent workflows.

GRE Vocab Master

apps.apple.com/app/id6758345755 · github.com/soojjung/gre-vocab-master

Jan 2026 — Present

  • Released a 1,560-word GRE vocabulary learning app to web and iOS using React 19 and Capacitor, maintaining the product through App Store v1.5.
  • Fixed a silent quiz-generation bug involving inflected word forms through regex-based normalization and full simulation across 1,560 vocabulary items.
  • Integrated Sentry across web and iOS after a Supabase auto-pause incident, establishing production monitoring and reducing new crash detection time to under 3 days.
TypeScript React Capacitor Supabase Sentry Tailwind CSS Vite PWA

AI Policy Assistant for Newlyweds (bbumate)

bbumate.com · github.com/soojjung/bbumate-policy-chat

Aug 2025 — Nov 2025

  • Architected an end-to-end RAG pipeline using LangChain (LCEL) and ChromaDB, centralizing 113+ fragmented policies into a high-dimensional (4,096-dim) vector store.
  • Achieved a 66% reduction in system latency by implementing parallel document grading, reducing complex task processing times from 4.2s to 1.4s.
  • Engineered a fallback strategy (RAG → Query Re-write → Web Search) using the Tavily API, ensuring 100% query fulfillment even when domain-specific data was missing.
  • Developed a high-reliability grounding system with 97+ URL mappings, ensuring all AI responses include 100% source attribution to official documentation to prevent errors.
Python FastAPI LangChain ChromaDB

Health Supplement E-commerce Platform

github.com/soojjung/health-supplement-shop

Aug 2025 — Oct 2025

  • Built a Spring Boot + React e-commerce platform for health supplements, implementing core user flows across authentication, product browsing, cart, order, payment, refund, and review features.
  • Developed stable frontend authentication and order UX using React and TypeScript, including silent token refresh, route guards, and guest/member cart merging.
  • Implemented backend reliability patterns with Spring Security JWT authentication, Toss Payments approval/refund flows, order state management, optimistic locking, and EntityGraph-based N+1 query optimization.
Java Spring Boot Spring Security JPA MySQL React TypeScript Docker

Personal Color Self-Diagnosis Service

omct.web.app · github.com/soojjung/personal-color-self-diagnosis

Feb 2023 — Aug 2024

  • Engineered a 9-stage algorithm using 36 curated chips with a bonus round re-evaluating the top 4 candidates to resolve ties, ensuring a 100% diagnosis rate.
  • Scaled the platform to 83,000+ diagnoses and 27,600+ AAU while maintaining a 91.5% completion rate through an intuitive UI/UX.
  • Designed a custom tie-breaker logic and a state management pattern using Recoil atoms with localStorage persistence to handle complex diagnostic flows.
  • Achieved 83% organic traffic (30,000+ sessions) by ranking on the first page of Google for keywords through targeted SEO and SSR optimization.
TypeScript React Next.js Recoil Styled-components Firebase

Additional