Fullstack Engineer with experience in building API & MCP
First seen by Alion on Sep 18, 2026.
If you want to build the interface other people’s AI agents depend on, not just another dashboard, keep reading.
About GetHookd
GetHookd.ai is an ad intelligence platform used by 5,500+ eCommerce and marketing teams to find and create winning ad creatives. We maintain a searchable library of 100+ million ads across Meta and TikTok, adding tens of thousands daily through a scraper fleet that runs 24/7. Semantic search over the whole thing. A data moat that grows every hour we stay online.
In July we launched our API and MCP server. AI agents now query our platform programmatically, and it’s our fastest-growing usage channel. Some of our most active users have never seen our UI. They’re Claude, Cursor, and custom agents built by teams we’ve never spoken to.
GetHookd is built and run by Meliora, a product studio in Sofia with two Shopify app exits behind us. GetHookd is our current focus and our biggest bet. A startup team with a studio’s track record, profitable, no outside funding, allergic to bureaucracy.
The Role
Straight version: the API and MCP server shipped fast because they had to. They work, people use them, and usage is outgrowing the first version. We’re hiring you to own that surface and make it the best way to access ad intelligence anywhere, for humans and for agents.
You’ll work across the stack: API design, MCP tools, auth and rate limiting, docs, SDKs, and the parts of the web app where developers sign up, get keys, and understand what they’re paying for.
The stack: PHP, Laravel, Python, Node.js, React.js, TypeSense-powered semantic search, GCP (GKE) with Cloudflare at the edge, GitHub Actions, and a product team shipping daily.
What You’ll Own
- The public API: endpoint design, versioning, pagination, error contracts, etc.
- The MCP server: tool design, descriptions, schemas, and response shapes that an LLM can actually use correctly on the first try
- Documentation that’s accurate because it’s generated from or tested against the real thing
- Auth, API keys, scopes, rate limits, and usage metering tied to billing
- Observability for the API channel: who’s calling what, what fails, where agents get confused, and what that tells us about the product
- Fullstack features that support the platform:
- A feedback loop between API usage data and the product roadmap
Who We’re Looking For
- 5+ years building production web applications across backend and frontend
- You’ve designed and shipped a public or partner-facing API that other developers depended on, and you’ve lived with its mistakes
- Hands-on experience building MCP servers, LLM tool integrations, or agent-facing interfaces is a strong plus. If you’ve watched a model misread your tool description and fixed it, we want to hear about it.
- Strong opinions on REST, OpenAPI, and error design, loosely held
- Comfortable with auth flows, rate limiting, caching, and the unglamorous parts of running an API in production
- Experience with search-heavy products (TypeSense, Elasticsearch, OpenSearch) is a plus
- You write docs as part of the feature, not after it
- You use AI coding tools daily and know exactly where they fall over
- You design for the caller who will misuse your API, including an agent that takes your error messages literally. We will ask about this in the interview.
This Role Is Not For You If
- You want to stay on one side of the stack
- You think developer experience means a nice docs theme
- Breaking changes are someone else’s problem
- You need a spec before you can start. Often you’ll be writing it.
- You want to rewrite the API in month one. It has scars, but it has paying users.
What We Offer
- Hybrid: week 1 in-office, then Mon–Wed office / Thu–Fri remote
- 20+ days paid leave, growing with tenure
- Private health insurance and an optional sports card
- Real product ownership. No red tape. No fluff.
- A small, honest team that ships things that matter
How We Work
- No Black Boxes. Share context early, often, and publicly.
- Feed Forward. Feedback is fast, clear, and focused on what’s next.
- Own the Outcome. We finish the job, not the task.
- Hack a Better Way. Prototype, test, adapt.
- Flex & Flow. Plans change, we move.
The Process
Four steps, two weeks total:
- a short written screen (3 questions, 20 minutes)
- a technical conversation including a review of a real endpoint or MCP tool from our codebase
- a paid 3-hour practical
- and a founder conversation.
Happy to hear from you!