Full-stack & AI Systems Engineer building software, developer tools, and agentic systems.

I work across the full software lifecycle — from product architecture and distributed systems to AI-assisted development, automation, infrastructure, and developer experience.

My current focus is on building reliable agentic software systems and exploring how AI agents can become useful, controllable, and trustworthy participants in the software engineering process.

🌐 aayurtshrestha.com.np

I'm currently exploring the intersection of:

• 🤖 Agentic AI & Multi-Agent Systems
• 🧠 AI-assisted Software Engineering
• 🛠️ Developer Tools & MCP
• 🏗️ Software Architecture & Distributed Systems
• 🔬 Reliable & Trustworthy Generative AI
• 🌐 Full-stack Web Applications
• 📊 Data-driven Applications & Automation
• ☁️ Self-hosted AI Infrastructure

I particularly enjoy taking ideas from prototype → architecture → implementation → deployment → automation.

A design system and UI ecosystem inspired by the scale and character of the Himalayas.

The goal is to build reusable, opinionated UI primitives that can be shared across applications while maintaining a consistent design language.

Focus: React · TypeScript · Design Systems · shadcn/ui · MCP · Developer Experience

An MCP server designed to expose structured project and design-system knowledge to AI agents.

The broader idea is to move beyond repeatedly explaining project conventions to an agent and instead make those conventions discoverable and machine-readable.

Focus: MCP · AI Agents · Developer Tooling · Architecture Governance · AI-assisted Development

A data-driven platform for analyzing the Nepal Stock Exchange ecosystem.

The project combines market data collection, normalization, analytics, automation, and AI-assisted analysis into a single system.

Focus: Node.js · SQLite · APIs · Data Pipelines · Automation · Analytics

An event discovery and ticketing platform designed around the Nepalese community in the UK.

Built as a multi-tenant application with event management, ticketing, payments, and content management.

Focus: Next.js · Payload CMS · PostgreSQL · Stripe · Multi-tenancy · TypeScript

One of my main areas of exploration is agentic software engineering.

Rather than treating AI as simply a coding assistant, I'm interested in systems where multiple specialized agents can participate in a controlled engineering workflow.

A typical workflow I'm experimenting with looks like:

Human
  │
  ▼
Task / Intent
  │
  ▼
Agent Orchestrator
  │
  ├── Research
  ├── Planning
  ├── Implementation
  ├── Code Review
  ├── Testing
  ├── Architecture Validation
  └── Documentation
          │
          ▼
      Verified Change
          │
          ▼
        Human

The important part isn't simply making agents autonomous.

It's making them:

• observable
• constrained
• repeatable
• testable
• architecture-aware
• easy for humans to review

This is also why I'm interested in Model Context Protocol (MCP) and tools that allow agents to discover project-specific knowledge instead of relying entirely on prompts.

I'm particularly interested in the future of software engineering when AI agents become active participants in development.

Areas I'm exploring include:

How can we evaluate whether an AI agent actually produces software that satisfies engineering requirements rather than merely generating code that appears correct?

How can agents participate in:

• requirements analysis
• architecture
• implementation
• testing
• debugging
• code review
• documentation
• deployment

while keeping humans meaningfully involved?

I'm interested in systems that make AI-generated decisions and changes more:

• verifiable
• explainable
• reproducible
• auditable
• constrained

Rather than replacing developers, I’m interested in how humans and agents can work together through well-defined interfaces, tools, feedback loops, and governance.

TypeScript · JavaScript · Python · PHP · SQL

React · Next.js · Vue · HTML · CSS

Node.js · Express · GraphQL · REST APIs

PostgreSQL · MySQL · SQLite · Redis · MongoDB

LLMs · MCP · Agentic Workflows · Multi-Agent Systems · RAG · AI Automation

Docker · Linux · GitHub Actions · CI/CD · Nginx · Traefik

Software Architecture · Distributed Systems · Testing · API Design · System Design

I care about more than getting an application to work.

I generally think about systems in terms of:

Architecture
    ↓
Boundaries
    ↓
Contracts
    ↓
Implementation
    ↓
Testing
    ↓
Observability
    ↓
Automation
    ↓
Continuous Improvement

I prefer systems where important decisions are explicit and where tooling can help enforce engineering conventions.

This is one reason I'm increasingly interested in building developer tools and MCP servers rather than relying solely on documentation and prompts.

I'm gradually building an ecosystem of reusable tools around:

• AI-assisted development
• MCP
• design systems
• software architecture
• agent workflows
• developer experience

Some projects are application-focused, while others are experiments exploring how AI agents can interact with real software projects.

I have a background in Computer Science and Software Engineering, with professional experience building full-stack applications, APIs, distributed systems, automation pipelines, and production software.

I've worked across:

• product development
• recruitment technology
• event platforms
• data systems
• enterprise integrations
• API development
• cloud infrastructure
• AI-assisted applications

My engineering interests have increasingly moved toward the intersection of software engineering and artificial intelligence.

I'm currently focusing on a few connected themes:

          AI
           │
           ▼
     Agentic Systems
           │
           ▼
  Software Engineering
           │
     ┌─────┴─────┐
     ▼           ▼
 Reliability   Developer
               Tooling
     │           │
     └─────┬─────┘
           ▼
     Better Software

The long-term goal is to understand how we can build software systems where AI agents are not just powerful, but reliable engineering collaborators.

I use GitHub as both a place to ship production software and a laboratory for experimenting with new engineering ideas.

You'll find a mix of:

• production applications
• open-source tools
• MCP servers
• AI experiments
• infrastructure
• research prototypes
• developer tooling

Some repositories are polished products; others are deliberately experimental.

I'm interested in connecting with people working on:

AI Agents · Software Engineering · Developer Tools · MCP · Generative AI · Distributed Systems · Open Source · AI Research

🌐 Website: aayurtshrestha.com.np

💻 GitHub: github.com/aayurtshrestha

Build systems that don't just work — build systems that can be understood, verified, and improved.