I turn ideas into products people actually use, and I enjoy every step of the journey: the scrappy first prototype, the moment the flow finally clicks, and the hardening that makes it ready for the real world.
I care about software that is useful, fast, and a little memorable: clean architecture under the hood, a considered interface on top. My daily toolkit is Next.js, TypeScript, tRPC, Prisma/Drizzle, Postgres/MongoDB,MERN,Eve,Django,FastAPI etc, and I'm increasingly building AI agents while pushing into Rust, Go and blockchain.
- 💬 EasyChat Support: an AI-powered customer support platform for businesses in Nepal, built as production-grade software
- 🌿 Ojas: a second production-grade product, currently in active development
- 🗂️ Everything else: case studies and other projects live in my portfolio at sudarshandhakal.com.np
I treat AI agents like any other production system: they need structure, limits and care.
- 🧠 Memory: designing short- and long-term context so agents stay relevant and consistent across conversations
- 🛡️ Guardrails: validating inputs and outputs, and constraining what an agent can say and do
- 🔐 Security: least-privilege tool access, protected secrets, and defenses against prompt injection
- 🚀 Deployment: shipping agents to production and keeping them reliable, with LangChain and n8n in the toolbox
| Area | Examples |
|---|---|
| 🌐 Full-stack products | SaaS platforms, dashboards, admin panels, auth-protected apps |
| 🤖 AI agents & automation | Customer-support agents, LLM-powered assistants, workflow automations |
| 📈 Data & market tools | Screeners, leaderboards, real-time dashboards |
| 🎨 Interfaces | Responsive, polished UIs with intentional visual design |
| Focus | What I'm exploring |
|---|---|
| 🦀 Rust | Systems programming and high-performance backends |
| 🐹 Go | Fast, simple, concurrent services |
| ⛓️ Blockchain | Smart contracts with Solidity and Solana / Anchor |
| 💭 Jev | skips token-by-token generation for instant decisions for AI |
- Start small, ship early: get a working slice in front of real use quickly
- Clarity first: an interface should explain itself
- Production habits from day one: auth, validation, error handling and secure defaults, not afterthoughts
- Test the flow before overbuilding: let real usage decide what deserves more polish
- Keep experimenting: side projects and prototypes are how I level up
- Share resources: Blogs,notes,skill.md files directly from my portfolio
I like software that feels crafted: the small details that turn a tool that works into one that's a pleasure to use.
-
🎓 Pursuing a B.E. in Computer Engineering at National College of Engineering
-
🤝 Open to collaborating on web app & AI projects
-
💬 Ask me about TypeScript, React, Next.js, Node.js and AI agents
-
📫 Reach me on X or LinkedIn (badges above)
Languages
Web, Mobile and Data
AI Engineering & Production Infrastructure
Models and AI platforms
DevOps, Cloud & Observability