AI Research Engineer
LLM routing and platform engineering, LANTA inference workflows, model evaluation, and NLP/translation systems.
I build LLM systems, from routing and inference to the platforms that run them.

A policy-driven router, a centralized model gateway, and the infrastructure to observe and secure them.

Reproducible HPC workflows and a 4,683-case benchmark that makes quantization trade-offs visible.

A six-setting English-to-Thai evaluation across 2,785 WIPO descriptions and 45 classification classes.
I'm Sakchai, an AI Research Engineer at NECTEC in Thailand.
My work connects LLM evaluation and translation research with the infrastructure needed to run them: model gateways, HPC workflows, observability, and secure data services.
I've also built agentic business-intelligence workflows on top of ERP data and analytics pipelines. Across these projects, the common thread is making model behavior measurable and systems reproducible.
LLM routing and platform engineering, LANTA inference workflows, model evaluation, and NLP/translation systems.
Integrated four 1C ERP sources with Airflow and PostgreSQL; built idempotent ETL, reporting, and natural-language BI workflows.
Built Flask/Plotly dashboards and anomaly detection for 10+ traffic streams; automated reporting with a recorded 17× efficiency improvement.
Supported RAID, NAS, SAN, storage migration, and enterprise-storage infrastructure.
First Class Honors · GPA 3.8/4.0
Terminology-sensitive English-to-Thai translation with lexical and dense retrieval.
Read the case study ↗Top 10% · Super AI Thailand, 2024
Winner · Berkeley SkyDeck Hackathon, 2023
Fast & Efficient LLM Inference with vLLM · DeepLearning.AI, 2026