Marketplace Decision-Telemetry Architect — Cross-Border Commerce

The mandate

During traffic or partner disruption, the marketplace can see component latency but cannot explain whether a seller listing, fraud decision, delivery promise or refund has become unsafe. Teams create separate traces, business events and dashboards, leaving cross-border currency, customs and fulfilment fallbacks largely invisible. Leadership needs to know which customer promises remain supportable when one decision dependency degrades.

The deliverable is a Marketplace Decision-Telemetry Architecture covering listing publication, checkout eligibility, fulfilment promise and post-order remedy. It will define journey identity, context propagation, business states, service objectives, sampling, evidence retention, dependency maps and degraded-mode signals. A working reference view must reconstruct technical and human decisions without using individual telemetry for employee productivity scoring.

Milestone one in week three delivers four journey traces and the blind-dependency register. Week seven concludes milestone two with telemetry contracts, objective design and cost envelope. At week twelve, milestone three supplies reference instrumentation and injected-degradation results. The accepted architecture, operating playbooks, adoption sequence and trained review council constitute milestone four at week seventeen.

Acceptance requires operations staff to reconstruct twenty unseen customer and seller outcomes within fifteen minutes, identify the failed dependency and explain the fallback used. Injected payment, customs, inventory and carrier degradation must trigger the intended promise or remedy boundary. Engineering Finance must validate telemetry cost, and both sponsors must approve adoption after an internal team runs the exercise unaided.

The client will provide pseudonymised journey records, traces, business events, decision logs, service maps, customer policies, partner SLAs, incidents and telemetry cost. Product and operations owners will define accountable outcomes; client engineers implement reference instrumentation. Privacy will approve retention and access, while the CTO will resolve contested cardinality or signal priorities within two business days.

Why this is external work

SRE sees technical health, product teams see conversion and operations sees exceptions, but each optimises a different slice of the journey. Tool vendors propose more collection without pricing whether signals answer consequential decisions. External work can impose decision-level evidence, privacy boundaries and cost discipline while remaining independent of the observability stack eventually selected.

What you will own

• Trace listing, checkout, payment, customs, allocation, carrier and remedy decisions through technical services, partner dependencies and human intervention.
• Define common journey identity and context propagation without exposing unnecessary customer, seller or employee information in telemetry.
• Establish decision-level objectives that reveal when a component degradation invalidates eligibility, promise, protection or refund assumptions.
• Design sampling, cardinality, retention and aggregation by evidential consequence rather than collecting every possible attribute indefinitely.
• Run failures involving currency service, customs content, fraud dependency, inventory reservation and carrier update to test degraded behaviour.
• Quantify telemetry value and cost through faster reconstruction, avoided unsafe promises, reduced manual search and controlled data exposure.
• Transfer contract governance, scenario testing and signal-retirement decisions to a council spanning engineering, product, operations and privacy.

Candidate qualifications

• Architected end-to-end observability for a large marketplace, payment, travel or digital service with many external dependencies.
• Connected traces and business events to customer and partner decisions rather than limiting observability to service latency and errors.
• Designed graceful degradation where eligibility, promise or remedy changed as evidence quality or dependency availability deteriorated.
• Controlled high-cardinality telemetry cost and privacy without removing the context required for incident reconstruction.
• Ran cross-functional failure experiments involving partner systems and human operations under realistic transaction conditions.
• Delivered an implementation-neutral telemetry architecture and helped internal teams retire low-value signals after acceptance.

Non-negotiables

• The named architect must lead journey observation and degradation exercises in Istanbul with regional operations participation.
• No reseller, referral, implementation quota or investment interest may exist with observability, APM or data-platform suppliers considered.
• Telemetry cannot be repurposed for individual employee productivity assessment or retained beyond approved evidential need.
• Customer and seller identifiers must remain pseudonymised inside controlled client environments during the engagement.

1. 49 words maximum. Describe a customer decision that became unsafe while its supporting services individually met latency targets.
2. 49 words maximum. How would you control telemetry cardinality without losing evidence needed to reconstruct a cross-border order?
3. 49 words maximum. Which partner failure would you inject to test whether delivery promise degrades honestly?