I'm a freelance AI enthusiast with 18 years of experience in localization and multilingual translation QA, backed by a U.S. university degree. Though not a full-stack engineer, I’ve created working AI tools that solve real problems—proof that curiosity and persistence can go a long way. In 2024, I developed and deployed an AI-powered translation QA tool using Power Automate and Azure OpenAI—without prior experience with these tools—and reduced the manual QA workload by over 70% in real-world localization projects. The project sparked my interest in AI’s potential—especially in areas where language quality and evaluation intersect.
🛠 Featured Projects
🧩 RPA-Prompt-TranslationQA
An AI-powered QA tool that classifies translation errors and detects subtle meaning mismatches using prompt-engineered GPT evaluations inside a Power Automate flow.
- Automation: Structured error detection via two-stage GPT calls
- Output: Error severity, error type, back translation, and meaning alignment
- Tech: Power Automate, OpenAI API (GPT-4o via Azure), Excel Automation
🔎 Mistranslation-Detector
A tool that detects potential translation issues using GPT-based direct evaluation and back translation. By combining both approaches, it leverages GPT in a unique way to score and classify translation quality more effectively.
- Input: Excel translation pairs
- Output: Risk scores & explanation
- Tech: Python, OpenAI API (GPT-4o-mini), pandas
🤖 OAT-Chatbot
A Japanese-language chatbot that uses RAG (Retrieval-Augmented Generation) to answer questions based on private blog content.
- Search: Hybrid of vector search (ChromaDB) and keyword filtering
- Output: Natural responses grounded in blog context
- Tech: Python, Flask, OpenAI API, ChromaDB, janome, RAG
🪰 Skills
Prompt Engineering · Python · OpenAI API · Power Automate · RPA Integration · ChromaDB · Flask · GitHub · Excel Automation · RAG
📜 Certifications
Prompt Engineering Professional – Certified June 2025, demonstrating advanced prompt design and LLM evaluation skills.