Serverless Event-Backpressure Recovery Leader

The mandate

A high-demand ticket release triggered timeouts that caused clients and functions to retry, multiplying messages faster than downstream inventory locks and payment steps could drain them. Concurrency quotas protected some accounts while starving compensating workflows, leaving seats unavailable and customers uncertain. The event-platform director was dismissed after recovery relied on manual queue deletion without a reconstructable order history.

The interim must join in Paris within ten business days and lead for nine months. Recruitment for a permanent event-platform leader begins after one major release completes with the new pressure and reconciliation controls, expected in month four. The final five weeks allow the successor to chair a capacity rehearsal, a poison-event review and the last live release command.

Handover requires each event path to have bounded admission, retry, age, concurrency and dead-letter behaviour; inventory and payment effects to be idempotent and reconcilable; queue pressure to degrade journeys according to agreed commercial priority; three release-scale exercises to pass; and the permanent leader to accept throughput envelopes, exceptions, provider limits and incident authority.

The interim may stop releases, cap admission, reallocate concurrency, quarantine event classes, change retry and queue policy, redirect teams and approve up to EUR 12 million within the recovery allocation. Permanent hiring, customer compensation, ticket-allocation policy, provider replacement and decisions above EUR 2.5 million require council approval. The seat cannot alter event rights or payment-control obligations.

Consumer-interface redesign, event acquisition, marketing demand generation and replacement of unrelated synchronous services are explicitly out of scope. The mandate covers serverless concurrency, event admission, backpressure, retry, ordering and reconciliation, release observability, provider quota and the operating command joining platform with fulfilment.

Why this seat is open

The release showed that elastic execution could amplify demand rather than absorb it when retries and downstream limits were not governed as one flow. Dismissal left serverless, queue, inventory and payment owners optimising local health while the customer journey remained unresolved. Temporary authority is needed to impose pressure contracts, prove them during real releases and transfer permanent command before the next season.

What you will own

• Reconstruct requests, retries, invocations, queue age, concurrency, inventory locks, payment effects, compensations and customer outcomes.
• Define admission, backpressure, shedding, retry, timeout, ordering, deduplication and expiry contracts for each event class.
• Decide concurrency reservations and downstream capacity envelopes by purchase stage, event priority and irreversible effect.
• Establish poison-event quarantine, dead-letter reprocessing and reconciliation without deleting evidence or replaying customer actions unsafely.
• Install journey-level pressure signals joining request amplification, queue growth, processing lag, lock health and fulfilment completion.
• Command rehearsals involving flash demand, provider throttling, slow inventory, payment degradation, duplicate events and partial region loss.
• Transfer flow maps, measured limits, event policies, exception history, supplier actions and release command through successor-led operation.

Candidate qualifications

• Held senior operational authority for serverless and event-driven platforms exposed to extreme, scheduled demand peaks.
• Recovered retry storms, concurrency starvation, queue amplification or unsafe replay involving irreversible business actions.
• Designed end-to-end backpressure across clients, functions, brokers, databases and external payment or fulfilment dependencies.
• Made degradation and load-shedding decisions using customer and commercial consequence rather than component throughput alone.
• Ran release-scale failure exercises that tested poison events, delayed consumers, provider quotas and reconciliation.
• Handed a recovered event platform to permanent leadership through observed rehearsal and live peak command.

Non-negotiables

• Can start in Paris within ten business days and attend all three release rehearsals plus live-event commands.
• Will accept exclusive executive responsibility and continuous escalation availability during ticket releases.
• Brings serverless concurrency and event-backpressure recovery; conventional web scaling or queue administration alone is insufficient.
• Must disclose current relationships with cloud, messaging, ticketing, payment and observability providers.

1. 49 words maximum. State your earliest Paris start and the largest retry-amplified event flow you personally recovered.
2. 49 words maximum. Describe a concurrency limit that protected one service while preventing system recovery elsewhere.
3. 49 words maximum. Which signal would make you shed new ticket demand before deleting or replaying queued work?