Porames Aimlaong

Computer Engineering fresh grad who has been tinkering with AI since high school. I love turning AI models into practical, full-stack applications using LLMs, RAG, and Vector Databases. I've built various custom AI systems through hackathons and recently applied my skills in the real world during an internship at STelligence (LLM + OCR pipeline).

While I've played with AI for years, I just began diving into formal research and ML pipelines about a year ago. Eager to bring my prompt engineering, Python, and FastAPI skills to a collaborative engineering team.

AI Engineer & Data Engineer Intern - STelligence

(2026-01)

Built an AI system for verifying data in documents using LLMs, Prompt Engineering, OCR, and Image Processing together, achieving 85% accuracy. Acted as both AI Engineer and Data Engineer — preparing data and checking quality before feeding it into the AI system.

• Designed custom prompts and called Gemini API to verify, cross-check, and summarize text data in documents, integrated with an OCR pipeline.
• Used PyThaiNLP to fix and clean Thai text before sending it into the pipeline, ensuring data is correct and consistent.
• Checked data quality before real use, finding errors and mismatches early to reduce problems in later steps.

AI Researcher (Trainee) - AI Association of Thailand (AIAT)

(2025-01 - 2026-12)

Joined Super AI Engineer Season 5, mainly responsible for preparing large-scale data (MySQL, PostgreSQL, MongoDB with tens of millions of rows) and building an ML Pipeline from data extraction, cleaning, feature creation, all the way to model training. Submitted a research paper.

• Pulled tens of millions of rows from MySQL, PostgreSQL, and MongoDB using a Sort-Limit-Merge approach: sort first, query in small batches, then merge later. This let us pull data fast and in volume without crashing the server.
• Built a full pipeline from scratch: cleaned data, removed duplicates, caught bad timestamps, then created new features from usage logs and summarized them per user.
• Used 5 Machine Learning methods (K-Means, K-Medoids, GMM, Hierarchical Clustering, Kernel K-Means) with PLS-SEM to group learners based on model residuals, then picked the best method using Calinski-Harabasz Index.

Master of Engineering - Computer Engineering - Chiang Mai University (2026-01)

Bachelor of Engineering - Computer Engineering - Mahidol University (2022-01 - 2026-12)