Pakin Siyaratchakhun

Computer Engineering student (GPA 3.93/4.00) at Kasetsart University with enterprise internship experience as a Machine Learning Engineer at Charoen Pokphand Group (CP Group). Experienced in architecting end-to-end ML/NLP pipelines, LLM information extraction, multi-stage semantic retrieval, and workflow orchestration with Apache Airflow and Docker. Proven ability to translate complex corporate and financial disclosures into scalable analytical assets.

Machine Learning Engineer Intern — Corporate Risk Office (CRO) at Charoen Pokphand Group (CP Group) (2026-04 – 2026-09)

• Multi-Market Document Intelligence Pipeline: Built an automated document ingestion pipeline (Playwright, PyMuPDF) across 5 Asian exchanges (SET, CSI 300, HSI, SGX, HOSE); localized target risk sections from 300+ page annual reports via Table of Contents parsing, page offset calibration, and LLM fallback heuristics, filtering out >90% of irrelevant text.
• Production Airflow Orchestration & Data Quality: Migrated manual workflows to production pipelines on Apache Airflow and Docker Compose; built 3 sequential DAGs with state checkpointing and automated Data Quality Gates, parsing 1,000+ structured risk items across SET and CSI 300 firms with 100% source traceability (COSO ERM).
• Multi-Stage Semantic Retrieval & Subsumption Engine: Architected a hybrid alignment pipeline combining Bi-Encoder dense search (Top-15) and GPU Cross-Encoder reranking (Qwen3-Reranker-8B, Top-5); engineered a 4-tier Subsumption Engine with CoT reasoning and a FastAPI human-in-the-loop web audit platform for compliance review.
• Financial & ESG Intelligence Pipelines: Built automated ingestion pipelines extracting 3 financial statements and 6-year workforce/employment ESG data for 500+ SET-listed companies; bypassed web application firewalls (WAF) via Playwright for financial statement data pipeline and KPI extraction.

B.Eng. in Computer Engineering – Kasetsart University (2023 – 2027)