Software engineer · AI & full-stack development

Rachel Lin.

Amherst College CS graduate building practical software across web applications, AI workflows, and data analysis.

Rachel Lin

Interactive · AI assistant

Ask Rachel

Curious about my work or what I do outside of it? Pick a question or ask your own, and an AI assistant grounded in this portfolio will answer in my voice.

Rachel’s assistant

Hi! I’m Rachel’s portfolio assistant. Ask me about her projects, experience, skills, or what she does for fun.

Answers are generated by Google Gemini from the facts on this page and may be imperfect. For anything else, email me.

Experience

Work, volunteering & internships

Full Stack Developer

Keelung For a Walk Program· Remote · Keelung, Taiwan

  • Improving admin and user experiences in a Vue.js/TypeScript SPA with admin editor tools and multi-code structure support.
  • Building AR.js WebAR challenges, location-based interactions, completion pages with timer functionality, and multilingual map synchronization, supporting delivery of 200+ unique challenge codes.
  • Integrating client-side state persistence and Firebase deployment workflows.

Full Stack Developer

IHAI Time (Initiative)· Remote

  • Developing a volunteer management platform using React, FastAPI, and PostgreSQL to support shift coordination, approvals, and time-credit tracking, including automated token-credit issuance.
  • Designed backend services, REST APIs, and relational data models; developing AI-assisted spreadsheet imports to standardize volunteer data and reduce manual entry. Evaluated on synthetic datasets, reducing 100-row import time by approximately 97% with 98.1% field-level accuracy.

Product Marketing & Content Strategist

Freelance· Remote

  • Led client engagements with 40+ brands across technology, AI, and lifestyle, including approximately seven campaigns with Perplexity AI.
  • Translated business goals into campaigns generating 4M+ cumulative views, refining messaging using engagement and conversion metrics.

Data Mining Intern

Amherst College· Amherst, MA

  • Built a Selenium pipeline to collect and analyze 10,000+ tweets. Used BETO (Spanish BERT) to generate sentiment labels and compared Logistic Regression and Random Forest classifiers to analyze trends across Latin American feminist hashtags.
  • Created 15+ visualizations with pandas and seaborn to communicate engagement patterns and temporal trends.

Lead Data & IT Team Member

The Remedy Project· Remote

  • Managed backend operations of a NationBuilder action-page platform for 1,000+ community members, supporting supporter data workflows, CRM segmentation, donation pipelines, and campaign automation.
  • Collaborated with co-founders to improve website architecture and user experience with Jinja, HTML/CSS, and JavaScript, building dynamic landing pages and streamlining action-page conversions.

COSC-112 Teaching Assistant

Amherst College· Amherst, MA

  • Led weekly labs and 12+ office hours for approximately 40 students in Intro to Computer Science II, supporting AI concepts, data structures, and algorithms.

Research Fellow → Gregory Call Intern

Amherst College· Amherst, MA

  • Built a Python discrete-event simulation of Starbucks ordering across mobile, drive-thru, and in-person channels, modeling 25 million customer entries across 25 queue-threshold scenarios.
  • Analyzed customer switching and balking behavior using NumPy and pandas; visualized response-time and customer-retention tradeoffs with Matplotlib and Seaborn.

Selected projects

GitHub ↗

Oct 2026

FundFlow — Fund Holdings Pipeline

PythonPlaywrightAWS S3MongoDB AtlasFastAPIpytestGitHub ActionsAWS OIDC

Built an end-to-end pipeline processing 19,000+ holdings across 5 iShares ETFs. Preserves source reports in AWS S3 and verifies SHA-256 checksums and row-level validation before loading MongoDB Atlas. Orchestrates ingestion through a manually triggered GitHub Actions workflow with AWS OIDC authentication, automated tests, and run summaries. Implements batch upserts, duplicate prevention, and source-to-record traceability, with a FastAPI dashboard on Vercel for processing history, searchable holdings, and server-side filtering and pagination.

  • 19,000+ holdings
  • 5 iShares ETFs
  • Source-to-record traceability

Aug 2026

U.S. Hurricane Landfall Detector

PythonGeoPandasShapelyPyProjpytestStreamlit

Built an end-to-end Python geospatial analytics pipeline that parses NOAA HURDAT2 storm records, reconstructs tracks through geometric interpolation, and detects 488 hurricane-strength landfall candidates across 18 Atlantic and Gulf coastal states since 1900 with Shapely and GeoPandas. Interpolates landfall time and wind speed for analysis and reporting. Achieved 94.9% reference coverage across matching candidates, with 56 automated tests and GitHub Actions workflows for validated data updates and dashboard availability checks, plus automatic deployment to Streamlit.

  • 18 coastal states
  • 94.9% reference coverage
  • 488 landfall candidates
  • 56 automated tests

Jun 2026 · In progress

AllyText

PythonFastAPILangGraphPostgreSQLDocker

Developing a Telegram-based AI personal messaging assistant with typed tool execution, permission checks, and LangSmith tracing. Validated backend behavior with 163 automated tests covering APIs, database operations, MCP tool calls, and failure handling using mock data.

  • 163 automated tests

Apr 2026

anti-brainrot

ElectronReactTensorFlow.jsNode.js

Built an Electron and React desktop app that detects phone usage locally with TensorFlow.js (COCO-SSD) at approximately 100ms per inference. Designed multi-window IPC for behavioral intervention overlays and refined false-positive handling through informal testing with approximately 10 users.

  • ~100ms per inference
  • Local computer vision
More projects Photobooth+, AlgoVisualizer, F1, MammothEDU & UGC

MammothEDU

ReactOpenAI APITailwind

AI-powered education web app built at the HackMIT Hackathon, offering increased curriculum flexibility for students and educators. Built with React, OpenAI API, and Tailwind CSS.

UGC Portfolio

TypeScriptWeb DesignUGC

My personal UGC portfolio, designed and built by me using TypeScript. My content creation work gives me client-facing experience, from understanding brand needs to communicating ideas and delivering content.

Education & training

CodePath

Applications of AI Engineering (Honors) · AI Open Source Capstone

Amherst College

B.A. in Computer Science · GPA 3.74 / 4.0

Coursework in AI, machine learning, data mining, data science, data structures, algorithms, computer systems, statistics via modeling, linear algebra, and computing and the arts.

Yonsei University · CIEE Seoul

Study abroad · Seoul, South Korea · Spring 2024

Studied computer vision and brain, cognition, and daily life.

Awards

CodePath

Applications of AI Engineering · Completed with Honors

MLH CivicHacks

Third Place, Best Accessibility Hack

Technical skills

Languages
Python, Java, JavaScript, TypeScript, SQL, HTML5/CSS3, R
Frameworks
React, Vue.js, Svelte, FastAPI, Node.js, REST APIs
AI & data
Prompt Engineering, LLM Workflows, RAG Pipelines, ETL, NLP, MongoDB Atlas, pandas, NumPy, Matplotlib, Seaborn, Computer Vision
Tools
Git/GitHub, VS Code, Vercel, Selenium/Playwright, PostgreSQL, Firebase, GitHub Actions/CI/CD, Docker, AWS S3, AWS OIDC, pytest

Activities

Leadership
Leader of Food Justice Alliance (1 of 4 leaders, 2021 - 2025), Co-President of Chinese Students Association (2021 - 2024)
Member
Women in Computer Science (WiCs), AI in the Liberal Arts (AILA), Asian Students Association

Rachel Lin · Amherst College ’25

Built with Svelte & TypeScript & matcha.