Fine-tuning · Education AI
AI Feynman Kannada Tutor
Multi-stage fine-tuning pipeline creating a reasoning-first physics tutor in Kannada — combining SFT and RAG for intuitive, grounded explanations.
- Multi-stage SFT: language → domain → grounding
- LLM-as-judge evaluation on 0–5 scale
- RAG with physics knowledge base
- 4-model progression with measurable gains
- Dataset and models on HuggingFace
Open case study →
Fine-tuning · Creative AI
AI Sitcom Scriptwriter
Teaching an open-source LLM to write The Office — reasoning-first screenplay generation with on-brand humor, character voice, and multi-step setups.
- SFT on reasoning traces + screenplay pairs
- Reinforcement fine-tuning (RFT) with PPO
- LLM-as-judge with 8 weighted metrics
- 3-model progression: Base → SFT → RFT
- Dataset and models on HuggingFace
Open case study →
