Why this exists:
Mumzworld is the Middle East's largest e-commerce destination for everything: mother, baby, and child. We serve millions of parents from Dubai to Riyadh, and we're now rebuilding the company to be AI-native end to end.
That means agents handling real customer - Recommendation systems that follow a child's growth from pregnancy to pre-teen. Copilots for our merchandising, buying, and logistics teams. Tooling that turns our catalog and parenting content into something genuinely intelligent. We want builders who can sit inside this and move the needle fast. Whether you're writing the code or defining what gets built, you'll own production work that real mothers use.
Two tracks
Pick the one that matches how you work. If you do both well, tell us in your application.
Track A: AI Engineering Intern
You'll be shipping production AI systems. Expect to:
- Build and deploy agentic workflows using Claude Code, Cursor, or similar tooling
- Design RAG pipelines over product catalog, reviews, and parenting content
- Wire LLM features into real customer touchpoints across WhatsApp, web, and internal tools
- Work with vector stores, evals, prompt iteration, and production observability
- Own the full loop from prototype to deployed service
What we look for:
- 2 to 3 strong AI or LLM projects we can actually try or read (GitHub, live links, Loom)
- Strong Python, comfort with APIs, basic cloud fluency (AWS or GCP)
- Experience with at least one of: agent frameworks, RAG, fine-tuning, or evals
- A shipped over perfect mindset
Bonus: Hackathon wins, open-source contributions, or published writeups of real projects.
Track B: AI-Native Product Intern
This is not a traditional PM internship. You'll be an operator who uses AI as the default tool for everything.
Expect to:
- Define problems worth solving, then ship prototypes yourself using modern AI tooling before anyone writes production code
- Run experiments against real user behavior on the Mumzworld app and WhatsApp
- Work directly with engineering, design, and merchandising to spec and launch AI features
- Own metrics, write crisp product docs, and drive decisions with data
- Sit close to the customer, including shadowing support and talking to mothers directly
What we look for:
- Evidence that you build, not just write. Show us a shipped side project, an AI prototype you coded yourself, a community you run, a bot you deployed, or a product you led end to end
- Sharp written thinking. We'll ask you to write, a lot
- Fluency with modern AI tools as part of your everyday workflow
- Comfort with ambiguity and strong product intuition
Bonus: E-commerce, consumer, or marketplace experience. Arabic or any MENA context is a plus but not required.
Real problems you could own:
- The Mumz Assistant. An AI concierge that understands product safety, age-compatibility, and gifting context. Target: cut support tickets by 40 percent.
- WhatsApp Personal Shopper. A chat-first AI sales assistant that helps a busy mom find the right car seat, stroller, or formula in under two minutes.
- Smart Merchandising Copilot. Auto-generate and localize product descriptions in English and Arabic, at catalog scale, with quality guardrails.
- Returns Intelligence. A model that predicts high-risk returns before checkout so we can intervene, saving millions in reverse logistics.
How We Work With AI
- We're pragmatic about tools. Use whatever gets you to a working system fastest, whether that's the latest frontier models, open-source where it fits, or agent frameworks when they earn their place.
- What matters is that you're fluent with modern AI tooling as part of your daily workflow and can pick up new tools quickly. Our stack will evolve, and we expect you to help shape it.
How to apply:
Send us:
- Three things you've built. Links to repos, Loom walkthroughs, deployed demos, or internal tools. For each one, one line on what it does and one line on what you learned.
- Your pick. Track A (Engineering) or Track B (Product). Both is fine if you've got receipts.
- A short note. If I joined Mumzworld, the one AI feature I'd build to make a mother's life easier is and how I'd approach it.
Selection process
We prioritize builders, so the process is practical. Shortlisted candidates complete a scoped take-home challenge, roughly 3 to 5 hours, focused on a specific Mumzworld use case (for example, a Baby Shower Gift Finder). You'll then review your build with our technical lead and get direct feedback.
Stipend is calibrated to the level of work you demonstrated, not your years of experience.
What You'll Get
Ownership of real production work, not a training-wheels project. Code and product decisions that reach millions of parents across MENA. A fast path to a full-time AI or Product role at Mumzworld for the interns who show up and ship.
Why Mumzworld
Most internships give you tasks. We give you the chance to shape how an entire region's parents shop, decide, and discover.
You won't just learn AI here. You'll ship it.
