Real work. Real results. No mockups. Here's exactly what I've built and the impact it delivered.
Hitchhiker AI
Next.jsClaudeFigmaChatGPTPerplexity
Of course! This site was built with AI too! I used ChatGPT and Perplexity for design insights to create the design guide, drafted it in Figma, then brought it to life in Next.js using Claude Code Max. The site demonstrates the power of AI-assisted development, combining strategic design thinking with rapid technical execution.
The Challenge
Needed a professional portfolio site that showcases both marketing strategy and technical execution capabilities while demonstrating AI-powered development speed.
The Solution
Used ChatGPT and Perplexity to research design trends and create a comprehensive design guide. Prototyped in Figma, then built the entire site using Claude Code Max for rapid Next.js development with TypeScript and Tailwind CSS.
The Results
Launched full site in 3 days instead of 2-3 weeks
Achieved 95+ Lighthouse performance score
Created reusable component library
Demonstrated AI-powered development capabilities
⏱️ Timeline: Total 12 work hours from concept to production without approval process
App Publisher Finder
FlaskPythoniTunes APIGoogle Play API
A web tool that solves the pain of app marketing fraud detection. When you spot one suspicious app, it automatically finds the publisher and downloads their entire app portfolio—so you can block them all at once. Supports both App Store and Google Play with auto-detection.
The Challenge
App marketers waste hours hunting down fraudulent publishers one by one. When you find one bad app, you need to manually search for all their other apps to block them—mentally exhausting and time-consuming.
The Solution
Built a CSV-upload web tool that automatically detects store type (App Store vs Google Play), finds each app's publisher, and returns their complete app portfolio. No login required, completely free, processes apps in bulk.
The Results
Reduced fraud detection from hours to seconds
100% free tool with zero monetization
Auto-detects store type from app ID format
Built and launched in 2 days
⏱️ Timeline: Less than 2 hours from idea to production
Instagram Reels Analyzer
PythonPlaywrightWhisper APIChatGPT
Marketing research made smarter! I built an automated Instagram Reels crawler that collects view counts, extracts audio content, and generates AI-powered summaries of influencer content. This tool transforms hours of manual video review into automated insights, helping marketers quickly analyze trends and find the best reference content.
The Challenge
Needed to analyze hundreds of Instagram Reels for marketing research, but manually watching and summarizing each video was extremely time-consuming and inefficient for gathering competitive intelligence.
The Solution
Built an automated crawler using Playwright for Instagram automation, integrated Whisper API for audio transcription, and leveraged ChatGPT to generate concise content summaries. The system automatically collects influencer names, view counts, and creates searchable video summaries.
The Results
80% time reduction analyzing 100 Reels
Automated data extraction from influencer profiles
AI-powered summaries via audio transcription
Scalable research for marketing insights
⏱️ Timeline: Total 7 days from concept to production
TikTok Crawler
PythonAppiumAndroidData Extraction
A Python-based crawling tool designed for educational and research purposes that automatically collects TikTok hashtag search results. Using Appium for mobile app automation and natural user behavior simulation, it efficiently extracts video metadata and stores data in various formats.
The Challenge
Manually collecting data from TikTok for specific hashtag trend analysis is time-consuming and inefficient. There was a need for an automated solution that could track large volumes of hashtags while avoiding platform restrictions, with scalability and easy maintenance.
The Solution
Built an Appium-based mobile app automation framework to collect data directly from the TikTok app. Designed with a modular architecture (config, core, ui, data, simulation) to manage each function independently, and applied the Page Object pattern to separate UI elements from business logic. Implemented natural user behavior simulation to bypass bot detection, and supported multiple output formats (CSV/Excel/JSON) to ensure data analysis flexibility.
The Results
Single and multi-hashtag crawling support for scalability
Automated data extraction (title, username, likes, post date)
Modular code structure for improved maintainability
CSV/Excel/JSON format data storage support
Natural behavior simulation for stable crawling implementation
⏱️ Timeline: Total 14 days from concept to production