Been playing around with 3D on the web and ended up building this.
The scene was made in Blender, characters with Tripo3D, and everything was brought together on the web with Three.js + Astra.
It’s also a live multiplayer experience, people can join the same space, walk around, talk to each other and just hang out together.
Try it here:
https://aeronaut.dailiess.app/
Or join my room:
https://aeronaut.dailiess.app/r/ikanishakm/ai
I develop and manage databases using Elasticsearch, Neo4j, and Supabase to deliver fast search, powerful data relationships, and scalable backends. I design efficient data models, optimize performance, and ensure secure, reliable data access. I handle everything from setup and integration to deployment and ongoing optimization.
Built during my internship at SMC Insurance Brokers. It reads motor insurance policy PDFs from 5 insurers with OCR and fills the spreadsheet rows the team used to type by hand, about 100 policies a day. The demo shows each field highlighted on the policy as it lands in its row. The screenshot uses synthetic data only.
0
55
Built an AI-powered market analysis agent using Node.js and OpenAI that retrieves market data and financial news, analyzes the information through tool calling, and generates automated market reports. The system also supports fault-tolerant tool execution and automated email delivery, demonstrating practical AI agent orchestration and API integration.
An enterprise SaaS platform that lets FMCG companies plan, simulate, and optimize trade promotions. It gave 15+ enterprise clients the ability to model 200+ promotional scenarios monthly, run what-if analysis through a Promotion Calendar Optimizer and Simulator, and forecast the impact of promotions on revenue, volume, and profitability across $50M+ in annual promotional spend. My focus was architecting and building the Python/Django backend that powered the simulation and optimization engine.