Information TechnologyFull-TimeEntry-level(0-1 yr)
Job Description
Project A is ALX’s AI learning platform: LLM products built on hypotheses about how people learn and how AI can help. Some of those hypotheses are wrong; this role’s job is to find out which ones, fast. It centres on uncertainty reduction rather than product vision. You own the Experimentation-Harness as a function — instrument the AI products well enough to evaluate them, source learners for betas (the binding constraint on the whole programme), run the ongoing beta tests, and close the loop: data reviewed, improvements shipped, next PoC out the door. UX validation is captured alongside learning validation, so we know not just whether it teaches but whether people can use it.
Specific Responsibilities
Learner Supply & Instrumentation
Build a repeatable pipeline of beta learners, engaging the right internal teams to keep them flowing — the constraint that gates everything else, and it rewards hustle over process.
Instrument the AI products in partnership with the LLMOps Engineer — you instrument; they build the evals over what is captured.
The Experiment Loop
Own the experiment loop — a backlog of the team’s biggest uncertainties, experiments designed against them, cycle time measured and shrinking.
Close the loop: data reviewed, UX and learning validation captured, improvements shipped, and the next PoC out the door.
Requirements & Skills
Essential Skills
Product experimentation: End-to-end experience with hypothesis, instrumentation, and decisions on A/B testing or structured product testing on a live product.
Technical fluency: Ability to talk concretely about instrumenting a product, read AI-eval results, and collaborate effectively with engineers.
Operator ability: Ability to recruit and coordinate real users, run a beta programme, and wrangle stakeholders.
Qualifications
BA/BSc/HND degree.
Desirable: Hands-on eval or analytics skills; EdTech experience.
Essential Traits for Success
Scientific mindset: Understanding what an experiment can and can't conclude, with openness to failing hypotheses.
Proactive problem-solving and ability to source beta participants with limited resources.
