TL;DR

You build the systems that make our company run on autopilot and the AI features that make our apps smarter.

Half your job is inside the app: agents, AI features, evaluation systems, and prompt pipelines that ship to real users. The other half is inside the company: autonomous agents on Slack and WhatsApp, automated reporting, internal tools, and workflows that replace manual work across product, marketing, and ops.

You are a builder. You do not write specs and wait. You prototype, ship, measure, iterate. Make.com on Monday, Claude Agent SDK on Tuesday, a Python script on Wednesday, a production agent on Thursday. Whatever the job needs.

# About Univenn

At Univenn, we operate as an **AI native startup studio** building and scaling mobile apps used by millions globally.

You will be part of our AI App Studio, where product, growth, creative, engineering, and data work closely together. We move fast, test constantly, and care about building products that users love, use, and pay for.
We kept what helps, removed what slows us down, and focused the company around ownership, speed, and shipping.

For us, **AI is not a feature**. It is how the company runs. Every manual step is a candidate for automation. Every bottleneck is a target. We want a builder who sees the same way.

What you will bring

## Inside the company, 

- Autonomous agents running 24/7 on Slack, WhatsApp, or any messaging layer, taking action without waiting for tickets

- OpenClaw class systems, autonomous Claude Code loops, and computer use agents that operate browsers, files, and APIs end to end

- Self running workflows: creative review, ASO updates, churn analysis, anomaly detection, agents that catch it, decide what to do, and do it

- Automated reporting pipelines pulling from Amplitude, RevenueCat, Adjust, Meta Ads, and App Store Connect, summarizing and acting on the data

- Internal tools that let non technical teammates direct AI workflows: creative brief generators, ASO assistants, user feedback synthesizers

- The glue layer connecting Notion, Slack, Linear, Make.com, Airtable, webhooks, and our app backends into one nervous system

- The validation loop: how humans approve, override, audit, and trust agent decisions without becoming the bottleneck

## Inside the app, AI features that ship to users:

- AI powered product features: chat interfaces, personalization, content generation, smart onboarding

- Prompt pipelines and agent workflows that run in production

- Evaluation systems: how do we know the AI feature is actually working? You design the answer.

- A/B testing infrastructure for models, prompts, and configurations

- LLM orchestration across providers: OpenAI, Anthropic, Gemini, open source via Groq, Together, Replicate

You decide what to build next based on impact, not because someone filed a ticket.

# What We're Looking For
 

- **Agentic mindset.** You think in autonomous loops, not single prompts. You have designed at least one system where an agent decides, acts, and reports back, not just one where it answers a question.

- **Builder mindset.** You would rather have a working prototype tonight than a perfect plan next week. You ship, measure, iterate.

- **Real AI systems experience.** You have built and shipped AI powered systems, not just prompts, but real systems that real people use. Bonus if some run without you watching.

- **Hands on with LLM APIs.** OpenAI, Anthropic, Gemini. You know what tool calling means and have debugged a flaky agent at 2am.

- **Fluent in AI coding agents.** Claude Code, Cursor, Codex, OpenCode. This is how you work, not something you tried once.

- **Programming fluency.** Comfortable with at least one real language, Python, TypeScript/JS, or similar, enough to write, debug, and deploy production code.

- **No code/low code fluency.** Make.com, n8n, Zapier, Airtable, webhooks. You know when to reach for code and when low code wins.

- **Integration speed.** You can read API docs and ship an integration the same day.

- **English fluency.** We work and communicate in English.

# Nice to Have

- An agent or automation you have built that is actively used by you or a team

- Experience with agent frameworks: Claude Agent SDK, LangGraph, Pydantic AI, Mastra, OpenAI Agents SDK, AutoGen

- Familiarity with computer use agents and browser automation: Browserbase, Playwright, Anthropic computer use, OpenClaw style systems

- Familiarity with diffusion models and image/video pipelines: ComfyUI, Replicate, Fal, Runway, Kling, Veo, Sora

- Experience with eval frameworks: Braintrust, Langfuse, custom harnesses

- Mobile app context: RevenueCat, Adjust, App Store Connect APIs

- Side projects that actually work

# This Role Is Not For You If

- You need a Jira ticket to start working

- You want to architect before you build

- You are more interested in research than shipping

- You think **AI engineer** means **I write good prompts**

- The idea of agents acting without human approval makes you anxious instead of curious

# What You Get

- Full latitude to decide what to build next, as long as it has impact

- Real production usage: what you ship will be used by teammates today and users tomorrow

- Direct line to the founder, who reviews and uses what you build

- Budget for any AI tool, API, or service you need

- A studio that treats AI as how we work, not a buzzword

- İzmir HQ, remote friendly for the right candidate