You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
🚀 Building multimodal AI products from idea → architecture → working systems
🧠 15+ years of enterprise AI & data delivery
🤖 Deep focus on Agentic AI, RAG, Local LLMs, and AI Data Infrastructure
🎙 Creator at SkillWisor – AI & career education on YouTube
📍 Based in Bangalore, India
🔭 Currently building: Vyonix Studio · Klassbook AI · AI Hubzy · iJEMS · Konvagence · OpenSkela
🌱 Always learning: MCP · Google ADK · LangGraph · Multimodal AI
AI financial coach – tracks expenses, habits & mood, then predicts your next move
🔒 Pre-launch
—
🛒 AI Hubzy
Curated AI tools hub
🔒 Pre-launch
—
💎 iJEMS
Desktop ERP for jewellery
🔒 Pre-launch
—
🤖 Konvagence
AI Agent Marketplace
🔨 In Development
—
⚙️ OpenSkela
TypeScript Agent OS (11,500+ lines)
🔨 In Development
—
🌠 Vyom (व्योम)
Multimodal LLM from scratch
🔬 Research
—
🎓 SkillWisor
AI education platform
🟢 Live
—
🔮 Finozzo AI
Your AI financial coach — more than a budget app
Expense tracking, budget guardrails, habit building, and mood journaling —
all connected by a brain that learns you.
Finozzo doesn’t just record what you spent. It correlates your mood, habits, and spending
to uncover hidden patterns. It predicts the best days and weeks for major purchases,
warns you when your mood might lead to impulsive spending, and even calculates your
financial stress index in real time.
Think of it as a financial therapist + AI data scientist living in your pocket.
Voice‑first, privacy‑first, and always one shake away.
⚠️ Mascot image is a placeholder. Replace the src with your actual image URL
(e.g., from a public repo or image hosting service) to show your mascot.
Build AI systems that are practical, intelligent, and usable in real‑world products – while helping professionals leverage modern AI for innovation and growth.
"Building the future of AI — one agent at a time."
TinyLLM with Attention Residue Architecture - PyTorch implementation of Attention Residuals (AttnRes) — arXiv:2603.15031 by Kimi Team. Replaces fixed residual connections with learned depth-wise at…
Learn how ChatGPT actually works by building one yourself — tokenizer → embeddings → attention → transformer → generation. Step-by-step with visualizations at every stage.