I work on problems at the intersection of machine learning, uncertainty quantification, and continual learning.
| Project | Description | Stack |
|---|---|---|
| BudAi | Natural-language AI workflow orchestrator for search, calls, and calendar booking | JavaScript |
| aegis-env | OpenEnv-compliant RL environment for AI safety incident response training | Python |
| Devsage | AI agent framework with rules, duties, memory, and workflow definitions | Markdown, YAML (spec-driven) |
| gladiator-arena | Fully on-chain gladiator fighting game with NFT ownership, verifiable combat via Chainlink VRF, and a decentralized marketplace | C++, Solidity |
| NexusArena | Full scale AI based solution for seamless experience in sporting events | TypeScript |
| RetinAI | Hybrid AI system for automated retinal disease classification (YayaBud collaboration) | TypeScript |
Strengths:
- 🐍 Python development
- 🤖 Artificial Intelligence & Machine Learning
- 🌐 Full-stack Web Development
- ⚙️ Scalable API design
- 🛢️ Databases (PostgreSQL, MongoDB)
- Uncertainty-aware AI systems and epistemic calibration
- Continual learning and the stability-plasticity dilemma
- Agentic AI safety and containment
- Human-AI interaction at the epistemic level
The Ego Gate — A mathematical framework for selective learning in AGI systems. Formalizes doubt (predictive entropy), curiosity (KL-divergence), arrogance (principled discarding), and dreaming (structured experience replay) as first-class information-theoretic operations governing stimulus routing and weight consolidation.