Customize Your Existing AI/Web Pipeline for Autonomous Vendure Debugging
Budget / Salary€30–250
TypeFreelance project
LocationRemote
Posted4 days ago
Customize Your Existing AI/Web Pipeline for Autonomous Vendure Debugging
Hello,
We are looking for an engineer with Vendure or a closely related Node.js/TypeScript e-commerce background to adapt the AI programming, website-debugging, testing, or automation pipeline you already use into a fully automated debugging, testing, and delivery-verification pipeline for a Vendure storefront. You do not need to invent a new framework from scratch: if you have a pipeline you already use for your own projects, you may fine-tune it, assemble it with other components, or replace our suggested components.
We already have a working pipeline prototype and a basic engineering foundation. The current design uses CAO as the coordination and supervision layer, OpenHands or CrewAI as bounded code-execution and collaboration workers, and Buzz for model or backend routing. The intended flow is:
Natural-language objective -> task decomposition and planning -> bounded code changes -> debugging and testing -> browser-based E2E verification -> necessary load/security checks -> reproducible delivery evidence and rollback information.
The existing framework already provides part of the coordination, code-execution, and evidence-collection foundation. We need you to complete the technical-review layer for genuinely unattended, end-to-end website debugging: identify missing decisions, tests, failure handling, and acceptance gates, then implement them as a repeatable automation process.
Relevant technical areas include Node.js, TypeScript, NestJS, GraphQL, TypeORM, PostgreSQL, Next.js, React, Redis, BullMQ or equivalent job queues, Docker, Git/GitHub, CI/CD, browser E2E testing, automated regression, multi-agent orchestration, tool use, bounded retries, permissions, observability, load testing, and authorized security or red-team checks.
The acceptance case is a redacted Vendure secondary-development migration. Acceptance requires reproducible E2E verification, a load-testing report, an authorized red-team testing report, clear failure evidence, a change summary, runtime identity, and rollback information.
You may use your own AI programming, debugging, testing, or automation pipeline and are not required to follow our architecture. OpenHands, CrewAI, or other mature open-source tools are acceptable, but your own established workflow is equally welcome. We prefer free open-source software and no additional software or service costs for the client.
This project is intentionally reuse-friendly for the contractor. We do not require exclusive ownership of a general-purpose pipeline, reusable components, general-purpose code, or your pre-existing know-how. After the project passes acceptance and payment is completed, you may retain, use, further adapt, license, or commercialize the reusable pipeline and general-purpose components developed for this engagement. We only need the agreed delivered version, documentation, and project-specific acceptance evidence for our own use. Our private code, client data, credentials, pre-redaction materials, and other confidential information are excluded from this reuse permission, and third-party open-source licenses still apply.
Expected outcome:
1. Accept a concrete Vendure debugging, migration, or feature objective.
2. Inspect the code and runtime, plan the work, and execute bounded changes.
3. Run appropriate unit, integration, browser E2E, and necessary load or security checks.
4. Preserve evidence and stop safely on failure instead of claiming success.
5. Produce reviewable test results, change summaries, runtime identity, and rollback information.
6. Leave humans primarily responsible for final acceptance, not step-by-step operation.
This is a small, focused first implementation with a target timeline of approximately 7 days. The goal is to adapt proven capabilities to a defined acceptance case, not to build a general platform from zero. Please provide a one-time fixed-price quotation, expected delivery timeline, brief technical proposal, relevant experience, recommended tools or replacement options, and main risks.
The public project briefing can be shared after initial discussion. Private repositories, legacy code, test materials, credentials, and internal contracts will not be shared publicly.
Public project briefing: https://github.com/vendure-ai-factory/vendure-project-briefing
Hello,
We are looking for an engineer with Vendure or a closely related Node.js/TypeScript e-commerce background to adapt the AI programming, website-debugging, testing, or automation pipeline you already use into a fully automated debugging, testing, and delivery-verification pipeline for a Vendure storefront. You do not need to invent a new framework from scratch: if you have a pipeline you already use for your own projects, you may fine-tune it, assemble it with other components, or replace our suggested components.
We already have a working pipeline prototype and a basic engineering foundation. The current design uses CAO as the coordination and supervision layer, OpenHands or CrewAI as bounded code-execution and collaboration workers, and Buzz for model or backend routing. The intended flow is:
Natural-language objective -> task decomposition and planning -> bounded code changes -> debugging and testing -> browser-based E2E verification -> necessary load/security checks -> reproducible delivery evidence and rollback information.
The existing framework already provides part of the coordination, code-execution, and evidence-collection foundation. We need you to complete the technical-review layer for genuinely unattended, end-to-end website debugging: identify missing decisions, tests, failure handling, and acceptance gates, then implement them as a repeatable automation process.
Relevant technical areas include Node.js, TypeScript, NestJS, GraphQL, TypeORM, PostgreSQL, Next.js, React, Redis, BullMQ or equivalent job queues, Docker, Git/GitHub, CI/CD, browser E2E testing, automated regression, multi-agent orchestration, tool use, bounded retries, permissions, observability, load testing, and authorized security or red-team checks.
The acceptance case is a redacted Vendure secondary-development migration. Acceptance requires reproducible E2E verification, a load-testing report, an authorized red-team testing report, clear failure evidence, a change summary, runtime identity, and rollback information.
You may use your own AI programming, debugging, testing, or automation pipeline and are not required to follow our architecture. OpenHands, CrewAI, or other mature open-source tools are acceptable, but your own established workflow is equally welcome. We prefer free open-source software and no additional software or service costs for the client.
This project is intentionally reuse-friendly for the contractor. We do not require exclusive ownership of a general-purpose pipeline, reusable components, general-purpose code, or your pre-existing know-how. After the project passes acceptance and payment is completed, you may retain, use, further adapt, license, or commercialize the reusable pipeline and general-purpose components developed for this engagement. We only need the agreed delivered version, documentation, and project-specific acceptance evidence for our own use. Our private code, client data, credentials, pre-redaction materials, and other confidential information are excluded from this reuse permission, and third-party open-source licenses still apply.
Expected outcome:
1. Accept a concrete Vendure debugging, migration, or feature objective.
2. Inspect the code and runtime, plan the work, and execute bounded changes.
3. Run appropriate unit, integration, browser E2E, and necessary load or security checks.
4. Preserve evidence and stop safely on failure instead of claiming success.
5. Produce reviewable test results, change summaries, runtime identity, and rollback information.
6. Leave humans primarily responsible for final acceptance, not step-by-step operation.
This is a small, focused first implementation with a target timeline of approximately 7 days. The goal is to adapt proven capabilities to a defined acceptance case, not to build a general platform from zero. Please provide a one-time fixed-price quotation, expected delivery timeline, brief technical proposal, relevant experience, recommended tools or replacement options, and main risks.
The public project briefing can be shared after initial discussion. Private repositories, legacy code, test materials, credentials, and internal contracts will not be shared publicly.
Public project briefing: https://github.com/vendure-ai-factory/vendure-project-briefing
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.