Currently debugging both code and the mysteries of human learning π€π
What I'm actually doing: Building AI systems for education that don't just hallucinate math solutions. Working with multimodal models (text + vision + logdata + voice + probably some magic), reasoning systems that actually 'think' before they speak, and trying to make AI tutors that won't traumatize students.
π¬ Current experiments: Can we make DeepSeek R1 explain calculus without sounding like a textbook? Do transformers actually understand fractions or just really good at statistical cosplay? And my personal favorite - building AI that helps teachers instead of replacing them (revolutionary concept, I know).
Hot take: LLMs in 2025 are like really smart parrots with PhD-level pattern matching. The interesting part isn't what they know, it's how they can help humans learn what they don't know yet.
Python JavaScript TypeScript Swift React Vue.js Next.js Nuxt.js
Node.js Express.js Flask FastAPI PyTorch Hugging Face PostgreSQL
AWS Cloudflare Docker GitHub Actions Figma
and so on...
Making AI less artificial, one confused student at a time π
- π₯ Kaggle Competitions Master (Top 1% worldwide)
- π OpenAI Researcher Access Program Recipient
- π Multiple Publications in Educational Technology & AI