Part 0 · 1 chapters · ~12 min
The Role, Read Closely
The five-round process, the job description grouped into six clusters with what each line is testing, a table mapping every line to grounded evidence and to the Learna course that holds the depth, and the rule for stating your own contribution precisely.
1
The JD, line by line
The process, as described by the recruiter: (1) an initial discussion, now done; (2) a take-home technical AI challenge; (3) SAATHI, a self-awareness and behavioural assessment; (4) a technical interview with the Head of Frontend; (5) a final round with the CEO. Each later round tests a different cluster of the JD.
| JD line (abridged) | what it is testing | your evidence (grounded) | Learna depth |
|---|---|---|---|
| own money and trading journeys incl. edge cases, degraded states, failure recovery | production ownership of the unhappy path | the loan application state machine; the non-indebtedness letter's eligibility gate, async generation polling and rejection-code handling | Trust P3, P4, P7 |
| onboarding, auth, KYC, deposits, withdrawals, wallets, trading, portfolio | breadth across the money surfaces | loans and KYC verification at Moniepoint; payments and wallets at Ajocard; Paystack wallet flows in Ohlify | Trust P1-P6 |
| architecture, design systems, CI gates, a11y, observability, performance | raising the bar beyond your own tickets | shell and design-system work you described; gates you can name | Architecture, Design, Disciplines |
| AI daily: agents, code gen, test gen, PR critique, refactoring | AI across the whole loop | M: eval harness, retrieval A/B, mock mode, prompt hardening; your agentic workflow and published skills | AI-native P0-P6 |
| correctness, security, maintainability, taste | accountability for what agents produce | how you review agent output; what you never delegate | AI-native P3, P5 |
| delivery triad with Product and UX | collaboration and interaction fidelity | a design you shipped exactly, and one you changed | Design P6 |
| challenge unsafe, misleading, inaccessible, complex | saying no well | part 5 of this course | Design P5-P6, Disciplines P0-P3 |
| conversion, latency, error rates, crash-free | outcome language | the non-indebtedness letter's target: loan complaints 6.58% → ~1.97%; 249 manual issues a month removed | SRE P7, Disciplines P4-P5 |
| hybrid / WebView, real-device performance | mobile realities | Flutter work (Ohlify); low-end Android constraints | Architecture P6, Computers P6 |
| a principal on B2C at scale | scale experience | state your part of the scale precisely (see the note) | Big-company FE |
say exactly what you did
A staff-level interviewer probes ownership quickly. "A $1B product" or "20M users" describes the company's scale, not your contribution. State your part precisely: the module you refactored, the flow you owned end to end, the numbers you measured yourself. Then let the company's scale be context. Precise claims hold up under follow-up questions. Inflated ones fall apart at the second "how exactly?".
THE JD, READ CLOSELY
six clusters of lines in the job description, what each is testing, and the evidence that answers it
swipe the figure sideways, or tap expand for full screen
1/6
money journeys
Ownership of money journeys, including the unhappy paths: "own major slices of core money and trading journeys in production, including edge cases, degraded states and failure recovery, not only the happy path". Tests: have you shipped and run a money flow, and do you think about the states nobody draws? Evidence: the loans state machine and its error states; the Letter of Non-Indebtedness rejection-code handling; the Trust course.