Founding Engineer

Magic Safari; Cape Town; On-site; ZAR 600,000 - 1,200,000; Full time

Job summary

Magic Safari is seeking a founding engineer to work across supply, operations, sales and product, building AI-driven tools that turn briefs into priced itineraries and bookable proposals.

You'll design agent architectures, fix rate cards, and develop systems that scale with thousands of properties and dozens of integrations. If you thrive in ambiguity and aim for excellence, this is your chance to shape Africa’s travel tech.

Qualifications

• Strong fundamentals of software engineering, including data modeling, API design, testing and debugging.
• Ability to read and write clean, well-tested code and quickly learn new languages.
• Experience with building AI-assisted systems or working with frontier models.

Responsibilities

• Design agent architecture to support pricing and routing.
• Fix broken rate cards and ensure data accuracy.
• Integrate with reservation systems and content sources.
• Build internal tools to review and improve agent performance.
• Collaborate across supply, operations, sales and product teams.

Skills

Software engineering fundamentals

API design

Testing and debugging

Production systems

Clean code practices

AI literacy

Hands-on problem solving

Tools

CLAUDE Code

Codex

Cursor

Job description

Safari planning is still done by hand. A travel agent sends a brief to a DMC (destination management company: the wedding planner of the safari world). The DMC emails a dozen lodges, airlines and transfer operators to check availability, prices and routing, and days later a quote comes back. It's slow, error prone and expensive. Travel agents want to sell Africa, but plenty of them end up selling somewhere else because Africa is so much work to plan.

Magic Safari makes safari planning computable, so a trip that takes days to quote takes seconds. A travel agent on a call with a client can add two nights in the Okavango and have a price, three alternatives and a feasibility check before the call moves on. Behind that is a lot of software, data and AI agents.

As our founding engineer you'll work across the whole business: supply, operations, sales and product. The kinds of problems you'll be solving:

• Giving travel agents tools that turn a brief into a priced itinerary and a proposal their clients can book
• Running operations with agents once a trip is booked: supplier confirmations, payments, vouchers and all the chasing in between
• Connecting to thousands of properties across dozens of reservation and availability systems, most of which were never built to talk to anyone
• Collecting and maintaining content (rates, photos, fact sheets, operating rules) from websites, PDFs, spreadsheets and inboxes
• Building the tools our own team uses to review what the agents do, fix what's wrong and make each agent better next time

We're looking for someone exceptional, whatever their level of experience. You might have ten years behind you or two. Either way you're solid on the fundamentals. AI writes a lot of our code, which makes fundamentals matter more: someone has to know when the output is wrong.

Nobody is going to hand you well defined tasks or a clean spec. Much of your job is figuring out what to do and how to do it. Some days you design agent architecture. Some days you hand-fix a single broken rate card at midnight, because it has to be right. You have to want both.

Like the Navy Seals or the Royal Ballet, we're not for everyone. But for the right person this is an awesome opportunity to go all-in with an AI-native team of experienced founders changing the way travellers experience Africa.

You'll fit if you

Safari planning is still done by hand. A travel agent sends a brief to a DMC (destination management company: the wedding planner of the safari world). The DMC emails a dozen lodges, airlines and transfer operators to check availability, prices and routing, and days later a quote comes back. It's slow, error prone and expensive. Travel agents want to sell Africa, but plenty of them end up selling somewhere else because Africa is so much work to plan.

Magic Safari makes safari planning computable, so a trip that takes days to quote takes seconds. A travel agent on a call with a client can add two nights in the Okavango and have a price, three alternatives and a feasibility check before the call moves on. Behind that is a lot of software, data and AI agents.

As our founding engineer you'll work across the whole business: supply, operations, sales and product. The kinds of problems you'll be solving:

• Giving travel agents tools that turn a brief into a priced itinerary and a proposal their clients can book
• Running operations with agents once a trip is booked: supplier confirmations, payments, vouchers and all the chasing in between
• Connecting to thousands of properties across dozens of reservation and availability systems, most of which were never built to talk to anyone
• Collecting and maintaining content (rates, photos, fact sheets, operating rules) from websites, PDFs, spreadsheets and inboxes
• Building the tools our own team uses to review what the agents do, fix what's wrong and make each agent better next time

We're looking for someone exceptional, whatever their level of experience. You might have ten years behind you or two. Either way you're solid on the fundamentals. AI writes a lot of our code, which makes fundamentals matter more: someone has to know when the output is wrong.

Nobody is going to hand you well defined tasks or a clean spec. Much of your job is figuring out what to do and how to do it. Some days you design agent architecture. Some days you hand-fix a single broken rate card at midnight, because it has to be right. You have to want both.

Like the Navy Seals or the Royal Ballet, we're not for everyone. But for the right person this is an awesome opportunity to go all-in with an AI-native team of experienced founders changing the way travellers experience Africa.

You'll fit if you

• Are strong on the fundamentals of software engineering: data modelling, API design, testing, debugging and running systems in production
• Write clean, well-tested code and pick up new languages and systems quickly
• Already live in Claude Code, Codex or Cursor, and know the edges of what frontier models can and can't do
• Build for the AI, not just with it. You write the CLAUDE.md files, the skills and the guardrails that make a fleet of agents produce good work at scale
• Have taste. You can tell a well-built system from a fragile one, and good output from slop
• Are relentlessly resourceful. You don't mind going off the map and inventing solutions as you go
• Hold yourself to almost unreasonable standards
• Have a bias to action. When someone says next week, you ask why not today
• Have read the handbook, and it sounds like you
  

Don't apply if you

• Want your hand held. You'll be given goals, not detailed direction or someone checking your steps, and you need to run with them on your own
• Think some tasks are beneath you. Here the glamorous work and the grunt work go hand in hand
• Treat AI as something to "get up to speed on." You should already be among the best AI users you know
• Want a nine-to-five. This is a startup with unreasonable standards