Background
- Creators have no quick way to learn about their audience across hundreds of comments; YTCA automates that.
- Built with Python (FastAPI backend), HuggingFace Transformers, scikit-learn, SpaCy, and JavaScript (Chrome extension frontend).
Features
- Sentiment analysis via Twitter-trained RoBERTa and a sarcasm model (helinvian/english-sarcasm-detector).
- Thematic clustering: K-Means (K=3) on TF-IDF vectors groups comments into Content, Technical, and General categories.
- PCA cluster map: interactive scatter plot showing comment clusters in 2D, hoverable for comment previews.
- Keyword extraction via SpaCy POS tagging, with clickable filters for comments.
- PDF export: generates a full sentiment report with video thumbnail, stats, keywords, and top comments.
Background
- I've been playing League of Legends for 6+ years (2019–2025). I built this project combining my coding expertise with my game knowledge.
- Built with React.js (frontend), JavaScript (backend), and JSON (data storage).
Features
- Add champions to the board by clicking their icons.
- Remove champions by clicking their icons in the selection/ban slots.
- Toggle between Pick and Ban modes to simulate a full draft.
- Use Swap Side to instantly switch viewpoint between Blue and Red team.
- Find champions quickly using the Search Bar or Role Filter.
- Each champion shows a Recommendation Score that updates live during the draft.
Background
- A web app for quick data visualization — users upload data and the platform auto-generates editable, exportable graphs.
- Built with Node.js (backend), React (frontend), DeepSeek R1T2 Chimera (API), and Chart.js (API).
Features
- Auto-extracts data from uploaded files.
- Provides an editable table for the extracted info.
- Recommends suited graph types for the data.
- Summarizes key trends in the data.
- Allows modification of graph visuals and supports exporting in multiple formats.
Background
- YouTube rate limited Discord bots from streaming audio, causing many popular music bots to shut down.
- As a fun challenge, I built my own private music bot.
Features
- Responds to @zlbot or !!! command prefixes.
- Commands: join, leave, play, skip, and help.
- Streams YouTube audio when given a link.
- Implements a song queue for continuous playback.
Background
- With a family business that uses customer information storage software, I wanted to build one myself.
Features
- Tkinter-based GUI with core features: storing, retrieving, searching, and editing user information.
- Data is organized by date in structured files for easy access.
