Research engineer and builder working across AI agent infrastructure, Web3 systems, developer tooling, and cloud-native automation.
I like building products that make technical workflows safer and easier to operate: local AI agent control planes, self-hosted sandboxes, Web3 payment prototypes, and infrastructure-heavy developer tools.
- Building Stacy, a trust-first local control plane for assigning, supervising, and auditing AI coding agents.
- Building StacyVM, self-hosted execution infrastructure for AI agents, code runners, previews, and automation workflows.
- Prototyping AI-assisted Web3 payments for Shopify merchants, including source analysis, provider recommendations, integration generation, simulation, and compliance-aware safety boundaries.
- Working in Web3 research and engineering, with public LinkedIn context pointing to Rome Protocol.
| Area | Recent work |
|---|---|
| AI agent infrastructure | Local agent orchestration, run supervision, approval gates, budgets, logs, cancellation, and workspace isolation. |
| Secure execution | Self-hosted sandboxes with Docker, gVisor, Kata, Firecracker-oriented architecture, live previews, SDKs, and pool mode. |
| Web3 commerce | Shopify-aware Web3 payment integration workflows, provider adapters, webhook simulation, metadata sync, and reconciliation-gated fulfillment. |
| Cloud and DevOps | Kubernetes, Docker, CI/CD, Terraform-oriented infrastructure, GCP, AWS, Jenkins, Ansible, and observability tooling. |
| Smart contracts and protocols | Solidity, Vyper, blockchain integrations, DeFi experiments, and developer-facing Web3 education. |
A local cockpit for running Codex, Claude, and other CLI agents with visible ownership, live logs, budgets, approvals, workspace isolation, and stop controls.
- Local-first agent operations
- Human approval gates and risk surfaces
- Git worktree isolation per task
- Adapter model for Codex, Claude, Cursor, Gemini, OpenCode, Pi, and gateway-backed agents
- Public CLI and npm release workflow through
stacy-cli
Self-hosted execution infrastructure for AI agents and automation workflows.
- Single-binary sandbox server
- Python and TypeScript SDKs
- Docker-first local development path
- Provider model for stronger isolation levels
- Live preview support for apps running inside sandboxes
- REST API, CLI, templates, pool mode, and multi-worker architecture
A local prototype for testing AI-assisted Web3 payment implementation workflows for Shopify merchants.
- Shopify theme and app source ingestion
- Payment provider recommendation engine
- Generated Liquid snippets, app routes, provider adapters, env templates, and deployment docs
- Payment lifecycle simulator with signed webhook verification and Shopify metadata updates
- Deterministic offline AI fallback, with optional OpenAI-backed orchestration
Languages: TypeScript, JavaScript, Python, Go, Solidity, Vyper, Java
Frontend: React, Tailwind CSS, Vite, Framer Motion
Backend: Node.js, Express, REST APIs, Postgres, local-first services
AI systems: Agent orchestration, tool adapters, sandboxed execution, evaluation, run logs
Web3: Ethereum, Solana, DeFi, smart contracts, wallet/payment integrations
Infra: Docker, Kubernetes, CI/CD, Terraform, Jenkins, Ansible, GCP, AWS, Linux
Testing: Vitest, Playwright, smoke tests, release gates, integration checks
- I enjoy turning messy emerging-tech workflows into products people can actually operate.
- I care about trust boundaries: credentials, secrets, budgets, approvals, isolation, and audit trails.
- I write and contribute around Web3, DevOps, cloud, and developer tooling.
- I like collaborating with founders, engineers, open-source maintainers, and technical communities.
- GitHub: github.com/arpan-mondal
- LinkedIn: linkedin.com/in/arpanrome
- Email: accesstoarpan@gmail.com
- Website: megaverse.co.in