Book a strategy call
Back to case studies
Education Administration Programs, AI

SPARK: Creating Educational AI Infrastructure

How a Norwegian-British education intelligence startup engaged a fractional CTO and a Business Analyst.

Challenge

SPARK is building adaptive learning infrastructure for underserved contexts, which rules out most of what education technology normally assumes: a reliable connection, a recent device, a single curriculum, and a support structure around the learner. The word that matters in the company’s own description is infrastructure. SPARK is not building an app on top of decisions someone else has already made, it is making those decisions itself. At the point it came to us, the company knew exactly what it wanted to exist in the world, and had no technical function of any kind.

That produced two requests, neither of them for a traditional engineer. The first was a fractional CTO: ten years and more of distributed systems design, AI products taken from proof of concept through to production hardening, and experience leading engineering teams rather than only building, on TypeScript, Java and Python with Kubernetes and AWS. Fractional, because what SPARK needed at this stage was senior judgement a few days a month rather than full-time leadership and the obligation to keep someone busy. The second was a Business Analyst with twenty years across education, investment, and human resources, able to turn a mission into specified requirements and to hold the conversation between the people who know the subject and the people who will eventually build the system.

Most companies at this stage hire a developer and discover what they should have specified somewhere throughout the work. The screening reflected the opposite intention. The CTO was assessed on architecture owned through to production rather than handed over at the diagram, and specifically on early-stage work, because taking a proof of concept to something buildable is a different skill from running an established platform. The analyst was assessed on requirements that had survived contact with engineers rather than documentation produced for its own sake. The first match was confirmed 36 hours after the request opened.

2
Talents Secured
3+
Months Duration
36h
Time-to-hire

Outcome

You cannot hire engineers until you know what you are asking them to build, and a company with no technical function has no way to answer that from the inside. SPARK brought the two roles that define the work, not the ones that do it: someone to make the architectural decisions and someone to turn the mission into requirements an engineer could act on.

  • Two requests across technical leadership and analysis, with the first engagement confirmed in 36 hours.
  • Technical architecture and business requirements, which is the work that has to happen before any engineering is worth starting.
  • Both roles brought into a company with no technical function at all, rather than added to an existing team.
  • Three months of senior input without a permanent hire, on a fractional basis sized to what the stage actually needed.
I am really glad you are on board. There are few times in history when man has had similar chance to combine intelligence with heart and talent to create the architecture for tomorrow’s world. The work Yotewo is doing is essential.
Christophe Melius Darnley-Brede
Christophe Melius Darnley-Brede
Founder, Spark

A Strategic, Multi-Phase Hiring Approach

Phase: Request

Two requests, neither for an engineer: a fractional CTO on distributed systems architecture across TypeScript, Java and Python with Kubernetes and AWS, and a business analyst on process mapping, requirements specification, backlog ownership, and technical documentation.

Phase: Matching

The CTO candidates were screened on early-stage research and work taken through to production rather than established platforms maintained, and the analyst candidates on requirements that have managed communication with an early-stage engineering team in the education domain, with 2 of 6 shortlisted going through to interview and hired.

Phase: Interviews

Founder personally ran both interviews and made both decisions, with the first match confirmed 36 hours after the request opened.

Phase: Engagement

Both engagements opened on a part-time workload, visible daily hours in the internal tracker, and monthly reporting with a detailed breakdown of hour commitments.

Phase: Support

Across three months, the workload scaled up and down, with any changes to it or extra hours requested and approved in the account.