At [**Refine**](https://refinehq.ai) — a no-code [AI agent
platform](https://refinehq.ai) — we simplify lead enrichment, prospect
research, and CRM data cleanup for modern GTM teams. Unlike traditional tools
like Clay or Apollo, Refine lets users run prebuilt [AI enrichment
agents](https://refinehq.ai) without writing prompts, building workflows, or
touching code.
We’re scaling fast with a clear product-market fit. You’ll join early, work
directly with the founder, and help build infrastructure powering real-world
use cases in [sales automation](https://refinehq.ai), marketing ops, and
recruiting.
* * *
### 🔧 What You'll Work On
* **Multi-agent infrastructure** – Design how AI agents collaborate to enrich, research, and clean data
* **Data verification logic** – Score and deduplicate lead data across enrichment providers
* **Agent feedback loops** – Capture user input to help agents improve themselves
* **Browser automation** – Build tools using Playwright to extract live data from websites
* **Prompt systems** – Manage structured prompts and reliable output chaining
* **Task orchestration** – Automate sequential agent tasks with failovers and retries
* * *
### 🧠 About You
* 2+ years of experience in software development (ideally at a fast-paced startup)
* Strong Python skills (async, scraping, API design, background tasks)
* Familiarity with [LLMs](https://platform.openai.com) (OpenAI, Claude, Groq) and prompt engineering
* Comfortable shipping fast and making autonomous decisions
* Bonus: experience with enrichment APIs (e.g. [Apollo.io](https://www.apollo.io/), [Clearbit](https://clearbit.com/), Zeobounce)
* * *
### ✨ What Makes This Role Unique
* Early ownership in a growing [lead enrichment](https://refinehq.ai) product
* You ship real features used by [sales teams](https://refinehq.ai), marketers, and recruiters
* High-agency, async-friendly environment
* No meetings for the sake of meetings — just fast execution
* Flat structure. No layers. No red tape.
* * *
### 🛠️ Tech Stack
* **Backend:** Python, FastAPI, Postgres, Redis
* **Agents:** Langchain (optionally), native async agent infra
* **Automation:** Playwright, browser automation, API orchestration
* **LLMs:** OpenAI, Claude, Groq (model-agnostic)
* **Infrastructure:** Lightweight, fast, API-first