Behavior Shaping
Post-training, prompting, and feedback changing model behavior.
Pittawat Taveekitworachai
I study how language models acquire
and express behavior.
Focused on reasoning models, post-training, evaluation, and agentic systems.
Reasoning Models Post-Training Evaluation Agentic Systems
Research Behavior, reasoning, and evaluation Engineering Systems, tooling, and deployment paths Design Readable interfaces for complex AI behavior
Explore my work
- Blog posts Essays, technical notes, and field observations on language models, agent workflows, and software practice. Explore 60 blog posts
- Publications Papers on prompting, reasoning, evaluation, game-related LLM work, and applied AI systems. Explore 39 publications
- Talks Conference talks, invited lectures, and workshops on reasoning models, evaluation, and AI engineering. Explore 29 talks
Research focus
Understanding and building reasoning systems.
My work studies model behavior, evaluation, and systems: how models reason, where they fail, and how we can shape and measure reasoning to build more capable and trustworthy AI.
Learn more about meEvaluation
Tests that expose failure modes before real-world use.
Reasoning Models
Open models and analyses for Thai, medical, and domain reasoning.
Agentic Systems
Tools and workflows that turn research into usable systems.
Latest writing
Field notes on AI systems
Essays on model behavior, evaluation, agentic workflows, and the engineering details behind reliable AI.
- AIEnglish edition
The Common Language of Knowledge: When AI Becomes the Lingua Franca of Understanding
9 min readAI is transcending the boundaries of linguistics, forging a 'common language' that unites disparate disciplines. As we enter a new era where questions matter more than answers.
- AI
The Silicon Mind: When AI Forces Us to Question Who We Are
6 min readAI promises a more efficient future for companies, but its endgame is often substitution rather than assistance. Once our work, voice, and behavior can be copied, we are pushed into a deeper question: what, exactly, makes a person a person?
- AIExternal
Rethinking How Medical AI Reasons: Introducing the Typhoon–SiData+ Ranked-List Medical Reasoning Model
1 min readA collaborative research project exploring whether small models can outperform frontier models—like Gemini 2.5 Pro—when trained to produce ranked lists that better reflect real clinical reasoning.
