Overview

Lead AI Scientist at SCB 10X, leading the Typhoon Team - Thailand’s leading AI research lab. Built and deployed production LLMs achieving 1M+ downloads and 30M+ API requests, with enterprise adoption at Toyota, PEA, and major hospitals.

Key Achievements

• Founding member of the Typhoon Team, building it from the ground up into Thailand’s leading AI research lab
• Led and launched multiple LLMs optimized for Thai, achieving over 1M downloads on Hugging Face and more than 30M API requests
• Contributed as both an individual researcher and supervisor across the majority of Typhoon Lab’s projects

Major Projects

• Typhoon Family of Multimodal LLMs – Developed one of the first text, vision, and audio multimodal models in Southeast Asia. Deployed by TDRI, VISAI, Siriraj Hospital, the Office of the Education Council, and others.
• Typhoon Audio2 – Among the first end-to-end speech LLMs in Southeast Asia.
• Typhoon T1 – The first reasoning model developed in Southeast Asia.
• Typhoon R1 – The most advanced reasoning LLM tailored for Thai, matching DeepSeek R1’s reasoning performance while surpassing it in Thai benchmarks.
• Typhoon OCR – A next-generation bilingual vision-language model for document parsing, outperforming GPT-4o and Gemini 2.5 Flash in Thai document understanding. Currently used by multiple enterprise clients including PEA and Toyota Leasing.
• Typhoon Translate – A lightweight Thai–English translation model optimized for edge devices, outperforming GPT-4o, Claude 3.7, and Gemini 2.5 Flash, with only 4B parameters enabling efficient edge deployment.
• Typhoon ASR Real-Time – A real-time Thai speech-to-text model with streaming capabilities, trained from scratch using a Transducer architecture. Delivers performance comparable to Google and Azure solutions at 156× lower cost.

Recognition

• Selected as one of AWS GAIA startups from over 1,000 applicants, receiving $1M in AWS credits
• Winner of the 2024 Techsauce Innovation Award

Overview

Senior AI & Software Engineer at SCB 10X. Main author of Typhoon LLM - the first competitive Thai Large Language Model. Established technical foundations that evolved into Typhoon Team (now Thailand’s leading AI lab).

Research & Development

• Typhoon LLM (Main Author): Developed Typhoon, the most competitive Thai Large Language Model at the time
• Evaluation Framework: Designed comprehensive Thai knowledge evaluation system from scratch for assessing LLM performance
• Continuous Pretraining Pipeline: Architected and implemented novel pipeline to adapt English-centric LLMs for Thai language
• Published Research: https://arxiv.org/abs/2312.13951

Strategic Contributions

• Established direction, themes, and frameworks for AI project guidelines across the entire Venture Builder team
• Collaborated cross-functionally on all Venture Builder AI projects and strategic initiatives
• Bridged research and product development for AI-driven startup ventures

Long-Term Impact

This foundational work established the technical and strategic groundwork for what evolved into Typhoon Team—now recognized as Thailand’s leading AI research lab for language model development in Thai and Southeast Asian languages.

Overview

Generative AI Engineer for a global B2C product (1M+ users). Built scalable LLM infrastructure handling 2K+ concurrent users and ~20 calls/sec. Achieved 10× cost reduction while maintaining production-grade performance.

Infrastructure & Scale

• High-Concurrency System: Built LLM infrastructure handling 2,000+ concurrent users with ~20 LLM calls per second
• Production Deployment: Deployed open-source LLMs at scale for real-time chat features
• Cost Optimization: Achieved 10x cost reduction while maintaining production-grade performance and reliability

Technical Engineering

• Designed and developed real-time LLM pipelines for conversational AI features
• Created innovative workarounds to overcome limitations of open-source LLM models
• Implemented efficient caching and batching techniques for optimized inference performance
• Built and maintained high-throughput infrastructure for sustained LLM operations

Core Expertise

Scaling challenges for real-time LLM applications, inference optimization, resource utilization, and production-grade deployment of open-source language models.

Overview

Tech Lead & AI Engineer at spacely.ai (SCB 10X venture builder). Led AI model development from concept to production. Architected end-to-end image generation pipelines and optimization techniques for production-ready system that enabled independent venture spin-off.

Technical Leadership

• Led AI Model Development: Directed the technical development of the image generation model from concept to production
• Production Pipeline Design: Architected and implemented end-to-end production-ready pipelines for AI image generation
• Scalability Planning: Established technical architecture and foundations to support platform scaling for independent venture launch

Engineering Contributions

• Developed optimization techniques to improve inference speed and enhance user experience
• Bridged technical implementation with business requirements and go-to-market strategy
• Brought image generation system to production-level quality within the project timeline

Outcome

Delivered production-ready AI image generation system and technical foundation, positioning spacely.ai for successful spin-off as an independent venture.

Overview

LLM Consultant at Deeple during early 2023 LLM revolution. Led Thai LLM prototyping, market opportunity assessment, and technical strategy for fine-tuning approaches. Developed comprehensive roadmap for organization’s LLM adoption.

Consulting Deliverables

• Thai LLM Prototyping: Led hands-on prototyping efforts for adapting large language models to Thai language use cases
• Market Opportunity Assessment: Evaluated and identified viable business opportunities for LLM applications in the Thai market
• Technical Strategy: Explored and recommended fine-tuning strategies for specialized domain applications

Strategic Impact

• Developed comprehensive roadmap for organization’s transition to LLM-powered solutions
• Defined technical requirements and implementation pathways for successful LLM adoption
• Positioned the organization to capitalize on emerging LLM opportunities in Thai language AI market

Overview

Founded and served as CTO of Kunana Company Limited, leading full-stack development and AI innovation across multiple commercial products and award-winning client projects.

International Recognition

Client projects developed under Kunana leadership won 11 international awards including:

• Cannes Lions 2023 (Save my voice project)
• Spikes Asia 2022 (Missing Person Reporter)
• Adfest 2022 (Missing Person Reporter, Visible-Track)
• MAD STARS 2021-2023 (Multiple projects)
• Adman Awards 2021-2023 (Multiple projects)

Commercial Products

Vmagine (AI-Art Generator App)

• Consumer mobile app for AI-powered image and art creation
• Developed proprietary generative AI model with automatic personalized fine-tuning system for AI avatars
• Featured full professional-grade image editor

KUU Creator

• Mobile application for online sellers to create advertising and content images
• Built in-house AI background removal technology and smart AI search system
• Featured by Bangkok Bank SME: https://www.bangkokbanksme.com/en/kunana-kuu-creator-application

QQ

• Enterprise questionnaire platform for analyzing open-ended responses at scale
• Leveraged Text Analytics and NLP AI for automated insights
• Partnership: Homerun Consulting

AI Eyes

• Enterprise AI solution for predicting human visual perception of packaging designs
• Built using Human Attention Prediction AI technology
• Partnership: Homerun Consulting

Award-Winning Client Projects

Save My Voice (2022)

• Mobile application helping hearing-impaired individuals train and visualize their voice
• Awards: Cannes Lions 2023, Adman 2023, MAD STARS 2023

Missing Person Reporter (2021)

• Social impact campaign leveraging face swap technology to aid in finding missing persons
• Awards: Spikes Asia 2022, Adfest 2022, Adman 2021

Visible-Track (2021)

• Accessibility application translating visual information to sound for visually impaired users
• Awards: Adman Award 2021, Adfest 2022, MAD STARS 2021

Overview

Frontend Developer at Allianz Technology. Improved critical application features by 10× through optimization. Led automation project for insurance formula verification across large-scale data processing systems.

Key Achievements

• Performance Optimization: Improved critical application feature speed by 10x through technical optimization
• Automation Project: Led special initiative to create automated solution for verifying insurance formulas within the application system
• Complex Data Interfaces: Developed intuitive user interfaces for handling complex insurance data and workflows
• Large-Scale Processing: Implemented performance optimizations for processing large datasets efficiently

Collaboration

Worked cross-functionally with business stakeholders, insurance experts, and development teams to align technical solutions with operational requirements and user needs.

Overview

Founding Software Engineer at Deeple. Built MVP and first production version of conversational AI chatbot platform. Developed in-house NLP models for Thai language that remained in production for multiple years (pre-LLM era).

Technical Contributions

• In-House NLP Model Development: Built proprietary NLP-AI model that remained in production for multiple years through the pre-LLM era
• MVP to Production: Led full development cycle from initial MVP to production-ready chatbot system
• Thai NLP Engineering: Designed and implemented natural language understanding components specifically for Thai language processing
• Full Stack: Developed the entire stack from frontend to backend, including database design and API development

Long-Term Impact

The NLP models and technical architecture developed as founding engineer proved highly robust, continuing to power Deeple’s production systems for multiple years after initial development.