Forward Deployed Engineer, live answer, requirement 9 of 12

Prototype to production

I take a prototype through generated tests and a two-approval merge gate to a production deploy that rolls back if the new pod keeps crashing or cannot start.

Built by Gus Assuncao of Gus IT LLC for a Forward Deployed Engineer application. Luca is the AI platform Gus IT builds and operates.

Try it

Describe a prototype. The judge scores six production gates from 1 to 5 with evidence, lists the gaps to close, and makes a ship, fix first or hold call by a fixed rule.

Public demo on Cloudflare Workers AI. Inputs are screened by Llama Guard 3 before any other model runs, requests are rate limited, and nothing here connects to a Luca or customer system.

The requirementDrive a solution from prototype or MVP, through testing, to a production deployment people depend on.

How it works

Build on the gatewayI build the prototype on the Luca Express model gateway, which calls several model providers. The gateway holds the provider keys, can fail over to another provider, and records the model, tokens and latency of each completed call.
Run the test suitesI run the platform test runner's eight fixed smoke suites, from infrastructure and authentication to data isolation and security. Its generators add endpoint checks for every app in the manifest list. They also add role permission, data isolation and input fuzzing cases across 22 platform domains.
Gate the mergeA scheduled job checks open pull requests every 15 minutes. It merges one only with at least two approvals from reviewers other than the author. It holds any pull request with a change request still open on its latest commit, or one GitHub reports as not mergeable.
Check the release recordBefore an image swap I run the release approval check by hand, because no deploy script calls it yet. It exits non-zero unless the commit has an approved release record that has not been deployed or rolled back.

Tools

Tool	What it does here	Status
Platform test runnerAvailable	Runs eight fixed smoke suites, endpoint checks for every app in the manifest list, and role, data isolation and fuzzing cases across 22 platform domains.	Available
Scheduled merge gateAvailable	Every 15 minutes, merges pull requests with at least two non-author approvals, no change request open on the latest commit, and a mergeable status from GitHub.	Available
Gateway deploy runnerAvailable	Swaps the gateway image behind a maintenance page and rolls back when the new pod keeps crashing, cannot pull its image, or has a bad container config.	Available
Release gates and rolling pathEarly access	The approval check exits non-zero without an approved, undeployed release record, but no deploy script calls it. Six in-pod smoke checks hold a failed release on the maintenance page. The rolling path is written but not live in production.	Early access
Readiness rubric demoDemo only	Scores a prototype you describe against six common production gates and returns ship, fix first or hold, on Cloudflare Workers AI.	Demo only

Available means shipped and in use. Early access means built but not generally available. Roadmap means designed, not built. Demo only means it powers the public demo, not customer deployments.

The write-up has every step of how I deliver this at a customer, with the tools and their status.

Read the write-up on gusit.de

The other answers

LLM applications in productionOn Luca, model calls normally go through a gateway that holds provider keys, records each chat call and sets a model per task.Prompt engineeringI write the output shape first, measure the prompt against each model's context window, and check replies against that shape.Retrieval-augmented generationChat answers over a customer's own documents, returned with the matched document names, after a check that the caller may use that persona.AI agents and orchestrationI build agents that call your systems through registered tools, still in early access, and workflows that start on business events.Request classification and routingI build automation that classifies each inbound request, plans bounded tasks, and holds high-risk tasks for approval; the dispatcher is in early access.AI workflow orchestrationI map the steps, API contracts, and data flows between a customer's systems and the models before I build the workflow.API and integration designI publish the API contract first, gate it with keys and scopes, then build integrations against it.Hybrid enterprise integrationI connect AI services to your identity, applications and data, and run the platform on Kubernetes on Azure Local.Approvals and exception handlingIn chat, I hold each Gatekeeper tool marked destructive until the signed-in user approves it, with expiry and a record. This path is in early access.Evals and output validationI built a versioned rubric judge for a tutoring app, and I trace failed or slow model calls to their provider.Client-facing deliveryI turn engineering work into updates that client architects, security leads and business owners can each act on.

Built by Gus Assuncao of Gus IT LLC for a Forward Deployed Engineer application. Luca is the AI platform Gus IT builds and operates. Contact gus@gusit.de. Claude is a product of Anthropic PBC. Cloudflare and Workers AI are trademarks of Cloudflare, Inc. Llama is a trademark of Meta Platforms, Inc. Other names are trademarks of their respective owners. Gus IT LLC is an independent engineering firm and is not affiliated with or endorsed by these companies.