Junior Software Engineer
We build an AI agent platform. Our agents talk to real customers over WhatsApp, Instagram, email and web chat, and connect to the systems businesses already run — CRMs, ERPs, calendars, payment providers.
We are a small engineering team with a written decision record, automated evaluation suites for every agent template, and a strong habit of shipping. A meaningful part of our codebase is written with AI coding tools and reviewed by humans — that is how we work, every day.
The stack you will work in
What you will do
Connect agents to real systems
Google Calendar and Sheets, Shopify, payment and shipping APIs, CRMs. Each one is a well-scoped, testable piece of work.
Configuration, prompts, evaluations
Write the YAML configuration, tool definitions and evaluation cases that keep an agent from regressing.
Ship across backend and console
Features and fixes in the Python backend and the TypeScript console, then take new customers live.
Read AI-generated code critically
Test it, question it, correct it. Working well with these tools means knowing where they are wrong.
What we are looking for
- Bachelor's degree in an engineering discipline (Computer, Software, Electrical-Electronics, Industrial or related) from a university. New graduates and candidates with up to ~2 years of experience are welcome.
- Solid fundamentals in Python or TypeScript, and the willingness to learn the other.
- Comfort with HTTP and REST — able to read an API document and write an integration against it.
- Hands-on experience with AI coding tools (Claude Code, Cursor, Copilot or similar) and genuine enthusiasm for working this way.
- Git fundamentals; readable commits and pull requests.
- A GitHub profile with code we can read. It does not need to be polished or popular — coursework, side projects and unfinished experiments all count. We want to see how you write, not what you have shipped.
- Ability to read technical documentation in English.
- Ownership: in a small team, "not my job" does not exist.
- Military service settled for the next two years — completed, exempt, deferred, or not applicable. We ask because the role needs an uninterrupted run, not to rule anyone out.
Nice to have: FastAPI, Celery, Redis or PostgreSQL experience · Next.js or React · Docker · anything you built with an LLM API · OAuth or webhook integrations · open-source contributions.
Working model
Hybrid: 2 days in the office, 3 days remote. The first two months are fully in-office, so you get the context, the codebase and the team properly.
You work on a MacBook Pro we provide, and the AI coding tools we use are paid for — they are part of the job, not something you fund yourself.
How the process runs
Three steps, no screening call. We read your application first — the GitHub profile and the repository you point us at — and the first conversation we have with you is a real one.
- A practical case study for the candidates we like: a task close to our real work, an integration against a public API. It will not eat your weekend.
- A one-to-one interview, face to face in the office, about the case you sent — your code, your decisions and how you worked with AI tools on it.
- Offer, quickly. No committee and no fourth round, so a strong candidate is not kept waiting for weeks.
Apply
Everything marked with * is required. Applications without a GitHub link are not reviewed.
Prefer email? Write to info@mangolab.ai with the same information — GitHub link included.
MANGO YAZILIM TEKNOLOJİLERİ LTD. ŞTİ. · AYDINEVLER MAH. SEDEF SK. D BLOK NO:1-3, MALTEPE, ISTANBUL · WE REPLY TO EVERY APPLICATION