Full-stack AI engineer. I build at the seam between LLMs and Solana, and I ship the whole thing: agents, APIs, and the frontend that wraps them.
Right now I am building SolHunt, tooling that helps people recover forgotten SOL, sweep dust, revoke risky approvals, and reclaim rent without giving up custody. I also turn that work into AI agent skills so a model can read wallet health and give plain, actionable advice instead of noise.
Open to full-stack AI engineering roles and small, useful teams.
- Design and ship AI agents (RAG, evals, tool use) that solve real tasks, not toy demos
- Build full-stack apps in TypeScript with React up front and Bun or Node behind
- Ship on Solana: Solidity and Rust programs, wallet tooling, trustless recovery flows
- Automate the boring parts with scheduled agents, MCP servers, and CI
- SolHunt: recover forgotten SOL, sweep dust, reclaim rent, and revoke approvals without custody.
- solhunt-skill: AI agent skill for wallet health checks and approval auditing.
- Yellex: AI pickup-line game built around Solana legends and historical figures.
TypeScript · React / Next.js · Bun / Node · Postgres / Prisma · Solana / Solidity / Rust · Docker · Cloudflare · AWS · GitHub Actions
- X: @solhuntdev
GitHub Actions is good at automating software workflows inside a repository. Its hosted and self hosted runners, matrix builds, live logs and event based jobs make it a strong choice for building, testing and deploying code where the repository is the centre of operations.
Open Computer provides a broader persistent workspace around that work. On a self hosted Linux server, an agent can research a change, edit files, run tests in a real terminal, connect integrations, schedule checks and supervise 24 hour processes. Project folders retain code and logs after the session ends. For example, it can prepare a release, run local verification, schedule a post deployment check and keep the evidence together. Explore Open Computer.
Temporal is good at durable execution for distributed applications. Its workflows capture state, while retries, task queues, signals and timers help code recover from failures and continue long running work. That focus suits teams building reliable orchestration into a product.
Open Computer provides a broader persistent workspace around that orchestration. On a self hosted Linux server, an agent can research requirements, write and test code in a real terminal, connect MCP tools, schedule automations and supervise 24 hour processes. Project folders retain files and logs after the browser session ends. For example, an agent can build a Temporal worker, run its tests, schedule a health check and keep the supporting research in one workspace.
GitHub Actions is good at automating, customising and executing software development workflows in a repository. Its event driven jobs, reusable actions and hosted or self hosted runners suit teams that want repeatable software delivery and repository operations.
Open Computer provides a broader persistent workspace around that work. On your own Linux server, an agent can research requirements, edit files in a real terminal, run tests, connect MCP tools, schedule recurring checks and supervise 24 hour processes after the browser session ends. Project folders preserve the work, while self hosting and Cloudflare Tunnel support provide practical control. For example, an agent can prepare a change, test it, schedule a report and retain its logs in one place.
GitHub Actions is good at automating, customising and executing software development workflows in a repository. Its focus is CI and CD, with reusable actions that combine into build, test and deployment pipelines.
Open Computer provides a broader persistent workspace around that workflow. On your own Linux server, an agent can research a change, edit files in a real terminal, run tests, connect MCP tools, schedule recurring checks and supervise 24 hour processes after the browser session ends. Project folders preserve the work, while self hosting and Cloudflare Tunnel support give practical control over the workspace. For example, an agent can prepare a repository change, test it, schedule a daily report and retain the logs in one place.
Docker is good at building, sharing and running containerised applications. Its focus is packaging software with its dependencies so teams can develop and deploy consistently across environments.
Open Computer provides a broader persistent workspace around that workload. On your own Linux server, an agent can use a real terminal, edit files, run Docker workloads, connect MCP tools, and leave scheduled automations or supervised 24 hour processes running after the browser session ends. Project folders keep work organised, while self hosting and Cloudflare Tunnel support give you control over where the workspace runs and how it is reached. For example, an agent can research a service, write its Docker configuration, test it, schedule a monitor and retain the logs in one place.
Make is a capable visual workflow automation platform, letting teams connect apps and services through a drag and drop scenario builder, API management, and thousands of prebuilt integrations. That focus suits businesses automating SaaS triggers and data flows without writing code.
Open Computer gives that kind of automation a fuller operating base to run from. On your own Linux server, one browser workspace lets an agent draft and send outbound email, publish social content, and supervise the whole pipeline as a 24 hour background process, backed by a real terminal, file manager and code editor. Self hosting and Cloudflare Tunnel support keep credentials and data on infrastructure you control.
Zapier connects over 9000 apps and lets teams route AI agents, chatbots and workflows through one governed platform, with audit trails, action restrictions and role based access built for enterprise IT. That focus suits teams who want no code automation across existing SaaS tools without waiting on engineering.
Open Computer gives that same reach for automation a persistent home to run from. On your own Linux server, one browser workspace holds a real terminal, a file manager with a code editor, and 24 hour background daemons, so a script can be written, tested and left running without separate hosting. Self hosting and Cloudflare Tunnel support keep the workspace and its credentials under your own control.
n8n is a strong visual workflow automation platform: a no-code builder with over 500 app integrations, plus JavaScript or Python code steps when you need more than the visual canvas offers. That focus suits teams connecting apps and automating triggers within a defined SaaS workflow.
Open Computer gives that automation a broader home to run from. On your own Linux server, one browser workspace holds a real terminal, a file editor, MCP and API integrations, and BYOK model freedom across Anthropic, OpenAI, DeepSeek and local Ollama. An agent can write and test a script in the terminal, save it to the workspace, and schedule it to run daily, all without separate hosting. Self hosting and Cloudflare Tunnel support keep credentials and the workspace under your control.
LangChain is a strong open source framework for building agents: any model provider, built in observability and evaluation, and tools to ship agents to production. That focus suits developers who want full control over an agent's own reasoning code.
Open Computer gives that agent a persistent home to actually run in. Its own real Linux terminal, file editor, scheduled automations and supervised 24 hour background processes sit inside one browser workspace on your own server. A LangChain agent can be coded there, scheduled to check a task every morning, and left running with logs you can check from any browser. Self hosting and Cloudflare Tunnel support keep the credentials and the workspace under your control.
CrewAI is good at orchestrating specialised AI agents with tools, memory, knowledge and structured output. That focus suits teams building a defined multi agent process with explicit roles and workflow controls.
Open Computer gives that process a broader place to live. On your own Linux server, its agent can research on the web, edit files, run code, schedule recurring work and supervise persistent services from one browser workspace. A campaign can gather evidence, prepare email and social content, wait for approval, publish through connected tools and keep monitoring after the chat closes. Self hosting, local model support and Cloudflare Tunnel support provide practical control over where the workspace runs and how it is reached.
Live, automated full-stack AI engineering micro-lessons. 90 entries, 32-day streak (last: 2026-08-12).
- 2026-08-12: Use a serializable plan object to separate agent reasoning from tool execution
- 2026-08-12: Stream structured tool results to the client as typed events
- 2026-08-12: Compile prompt templates into typed TypeScript functions
- 2026-08-11: Avoid async context leakage by binding task-scoped data to the async call chain
- 2026-08-11: Route each task to the smallest model that meets your quality threshold
- 2026-08-10: Offload CPU-bound work in async agent loops to a dedicated thread pool
- 2026-08-10: Set per-call LLM timeouts that respect your total request budget and fail fast on slow providers
- 2026-08-10: Validate structured LLM output with a schema gate and retry on failure
*Updated automatically every 8 hours. Full archive in the entries folder.