The setting
runs a global network between users and Internet applications. is its service that routes requests to AI models from different providers, for developers who build applications that use AI.
The production cutover protects launch timing or a complete explanation of every routing decision.
The assistant needs resilient provider switching, while audit reviewers need a record they can defend after an answer changes.
Who you’d be doing this for
“Risk asked why the answer changed, and I had three logs that told three different stories.”
Giorgos Ferrari · AI Platform Engineer · a character in this brief
Runs the customer’s shared inference platform and must answer engineering, finance, and risk questions about model behavior.
What is at stake
The customer cannot consistently explain which model answered a request after fallback. You must weigh the go-live date against a routing record risk reviewers can defend.
Why it isn’t already fixed
Every obvious fix costs something else. That’s the part you’d have to decide.
- go-live timing vs audit defensibility
- fallback resilience vs explainable behavior
- shared-wallet visibility vs routing flexibility
- automated telemetry vs verified evidence
Why Cloudflare
AI Gateway sits at the customer’s inference control point, where routing, fallback, logging, and unified AI usage meet.
Written with these in mind
Not your kind of problem? 100 more at Cloudflare, or browse every organization.
Where the organization context comes fromThe postings this brief was written from, and the reporting read before it.
The organization context comes from public information. The situation and characters are written for the brief.
Job postings it was written from
Public descriptions of the work at Cloudflare. They are listed as sources, not as openings. The lines under a posting are its own words, where it meets this brief.
- Senior Customer Engineer, Hong Kong (Pre-sales) (opens in a new tab)
- Leverage AI agents and internal platforms to automate routine tasks such as RFP/Security Questionnaire responses, sandbox provisioning, and basic telemetry monitoring, reclaiming your time for strategic customer advisory.
- Instead, you will operate as a quota-carrying technologist, utilizing AI-augmented workflows to eliminate administrative toil so you can focus on high-value architectural design and driving measurable business outcomes across the entire customer journey.
- Senior Customer Engineer, Named - LA (opens in a new tab)
- The Customer Engineer (CE) is the core driver of our technical go-to-market strategy: a trusted technical advisor and primary architect of the customer journey.
- Identify potential deployment risks or integration bottlenecks early, and provide remediation strategies to de-risk customer outcomes.
- Partner Engineer, Spain (opens in a new tab)
- Principal Partner Engineer, APAC (opens in a new tab)
- Partner Solutions Architect- Service Providers Location New York/ New Jersey, Dallas, Denver or Seattle. (opens in a new tab)
- Leverage Autonomous AI Tools: Champion the use of Cloudflare OS, AI tools (Agent PAT and Agent Echo) to automate standard, repeatable technical tasks, optimizing operational efficiency and freeing up bandwidth for high-touch, strategic global partner engagements.
- The Partner Solutions Architect (PSA) is a high-impact, strategic technical leader responsible for driving the comprehensive pre-sales and post-sales technical relationship with our top-tier Global and GEO Partners, including Service Providers (SPs) and Managed Service Providers (MSP)
Reporting read before this brief was written
The pieces closest to the situation in this brief.
Cloudflare released Auto Router in public beta for AI Gateway on September 30, 2026, adding automatic model selection and fallback between eligible models to the gateway layer instead of leaving routing to application code.
Cloudflare AI Gateway adds Auto Router for model routing · B2B LLM (opens in a new tab) · b2b-llm.com · Oct 1, 2026
This is the setup. The work is inside.
Running it puts you in the room: the full situation and its constraints, stakeholders who push back in their own words, and the decisions that are yours to make. What you produce becomes a Day One Plan — work you can show someone instead of describing.
Where similar work appears
Other live briefs connected to this one, and what connects them.
Similar problems at other organizations
Like this one, each of these is about an irreversible change forced by a fixed date.
Procurement closes in 21 days while risk cannot yet approve every production model route.
You must weigh a signed consumption path against promises the delivery evidence cannot support.
High-severity requests changed after a routing update, and the customer must report a material regression within seven days.
You must weigh continuity for business users against evidence that can withstand security review.
A failed regulated-container rebuild puts a contracted clinical launch at risk.
You must weigh the launch date against a production path the customer can defend after clinical use begins.
The same setting, from another role
Each is set around Cloudflare’s AI Gateway, written for a different role. They’re separate briefs, not the same scenario.
Discovery records are leaving sellers unsure which AI Gateway concern the buyer actually raised.
You have to decide where the extraction stops and the seller starts when a record's cost, access or routing concern is unclear, and who settles it before a next step goes out.
Prospects are hearing separate proposals for AI routing, security, and cost control.
You must decide what a partner can credibly lead and what stays in the direct technical conversation.
A $2 million platform agreement is stalled because the buyer cannot reconcile production AI traffic with its residency requirements.
You must weigh a credible multi-product commitment against terms that will survive technical and risk review.
Other problems with a similar purpose
In each of these, as in this one, the aim is to build machine-learning systems that earn trust by staying explainable where their errors carry real consequences.
An audit found sensitive fields in 7% of shared artifacts that manual review missed, and the classifiers disagree on the edge cases.
You have to build evidence for a production commitment you cannot easily reverse.
One threshold sends very different numbers of drivers into coaching in each region, and reviewers there agree less often.
You have to balance fair treatment of drivers against enough coverage to improve coaching.
Confirmed takeover messages are reaching guests, while legitimate pre-check-in threads are also being blocked.
You must weigh fraud prevention against a guest losing the instructions needed to enter a booked stay.
