At ALX, we’re unlocking the future we want to see. We’re catalyzing the transformation of Africa, by developing the next generation of bold, innovative, ethical and entrepreneurial leaders. We’re unlocking the potential of the world's largest workforce. The future is calling: be the answer.

Read more about this company

Role Summary

Project A is the AI layer of ALX’s learning platform: onboarding and profiling, a project guide that works alongside learners, the Project-Deconstructor, and the content mappers that connect it all to a competency model. These began as prototypes; the AI Engineer’s job is to make them reliable products. That means owning the systems learners actually touch, the context engineering that decides what an agent knows at any given moment, the tooling and service interfaces agents work through (such as the LLM’s access to the learner’s artifact window), and the day-to-day work of keeping long-running, multi-step agents reliable when real learners do unexpected things.

You will work in collaboration with Anthropic Engineers, a team of AI engineers, product managers and data scientists to design world class learning experiences.

Specific Responsibilities

Production AI Products

• Own Ai products in production, stable, observable, with regressions caught by evals before learners find them.
• Take the next prototype from working demo to maintained product, with evals built in from the start rather than bolted on.

Agent Architecture & Context Engineering

• Design the context and tooling architecture for the agents, what is in context, when, and why.
• Build and maintain the service interfaces agents work through, such as the application’s access to the learner’s artifact window.
• Keep long-running, multi-step agents reliable under real-world learner behaviour, with clear failure modes and recovery.
• Build the habit of testing what you ship — eval loops, regression checks, and fast iteration as a default way of working.

Skill Requirements - Essential

• Python & FastAPI: strong, production-grade experience with real users.
• Shipped LLM applications: at least one AI/LLM application you have shipped and can discuss in detail — what broke, how you found out, what you changed. Scale matters less than what you learned from its failures.
• Agent frameworks: working fluency with agent frameworks (we use LangGraph; equivalents fine) and a real point of view on context engineering.
• Testing discipline: the habit of testing what you build.
• AI-native coding with demonstrated harness engineering experience

Desirable (not required):

• React and full-stack range;
• Langfuse or similar observability; knowledge graphs.
• Educational/Edtech domain experience

Essential Traits for Success

• You build with AI, not just use it, proper harnesses, eval loops, and fast iteration.
• You can’t put an interesting problem down.
• You figure things out fast; we don’t need an exact-spec match if you have the base and the initiative.
• You favour non-standard evidence of skill, a portfolio of weird side projects beats a polished CV.

Person Specification/Attributes

• Courage: Willingness to speak up, challenge the status quo, and embrace new challenges.
• Humility: Openness to learning, seeking help when needed, and a focus on serving others.
• Adventure: A passion for setting ambitious goals, tackling difficult tasks, and finding joy in the journey.
• Initiative: Proactive problem-solving, a sense of ownership, and a willingness to go above and beyond.
• Resilience: The ability to bounce back from setbacks, persevere through challenges, and emerge stronger.

• Own Ai products in production, stable, observable, with regressions caught by evals before learners find them.
• Take the next prototype from working demo to maintained product, with evals built in from the start rather than bolted on.
• Design the context and tooling architecture for the agents, what is in context, when, and why.
• Build and maintain the service interfaces agents work through, such as the application’s access to the learner’s artifact window.
• Keep long-running, multi-step agents reliable under real-world learner behaviour, with clear failure modes and recovery.
• Build the habit of testing what you ship — eval loops, regression checks, and fast iteration as a default way of working.

• Python & FastAPI: strong, production-grade experience with real users.
• Shipped LLM applications: at least one AI/LLM application you have shipped and can discuss in detail — what broke, how you found out, what you changed. Scale matters less than what you learned from its failures.
• Agent frameworks: working fluency with agent frameworks (we use LangGraph; equivalents fine) and a real point of view on context engineering.
• Testing discipline: the habit of testing what you build.
• AI-native coding with demonstrated harness engineering experience
• React and full-stack range;
• Langfuse or similar observability; knowledge graphs.
• Educational/Edtech domain experience

• BA/BSc/HND