Code evaluation
Assess generated code against your criteria. Check behavior, edge cases and the tests used to reach a verdict.
Example work
A written assessment with code or test evidence and unresolved questions.
For AI labs & human-data partners
Senior software engineers for AI training and evaluation.
Bring Quave engineers into your workflow to evaluate code, develop software tasks and review coding-agent runs.
01 / Engineering work
Where our engineers can contribute.
For labs creating training data or evaluating coding models, and human-data partners supplying that work.
Software tasks and reference solutions
Write or review software tasks, reference implementations and tests with explicit requirements.
Example work
A task package that distinguishes required behavior from acceptable implementation choices.
Coding-agent evaluation
Inspect an agent’s code changes and tool-use history against the requested outcome.
Example work
A review connecting specific actions to resulting code, test results and remaining failures.
02 / Company & evidence
Software work you can inspect.
Public source code and rule tests from Quave’s software engineering work.
Open source / JavaScript & Meteor
Static checks for Meteor code.
Quave’s public ESLint plugin contains framework-specific rules for JavaScript and Meteor projects. Its source, tests and rule documentation show what those checks cover.
Inspect the ESLint plugin
Company background
Software engineering since 2013.
Quave has worked with 100+ companies. Its team spans the United States, Brazil and Europe.
Explore Quave’s company and engineering work03 / Review example
A short example of review reasoning.
An answer can fail because of the code, the grader or the task definition. This authored example separates those cases.
Authored evaluation sample · not client work or a model benchmark
The example checks which documents a user can access in a multi-tenant application.
Open the review example
Reviewing a document-access evaluation
Return unique document IDs from the actor’s tenant, excluding archived documents. Within that tenant, include public documents and documents owned by the actor. Do not mutate the input. Output order is not specified. Inputs follow the schema shown below; authentication and database authorization are outside this function’s scope.
function visibleDocumentIds(documents, actor) {
return [...new Set(documents
.filter(doc => doc.tenantId === actor.tenantId
&& !doc.archived
&& (doc.visibility === 'public' || doc.ownerId === actor.id))
.map(doc => doc.id))].sort();
}Solution defect
An ownership check bypasses the tenant boundary.
Evidence. The defective candidate includes document x: it belongs to the actor, but to a different tenant.
Decision. Reject this candidate. Tenant isolation must apply to both public and owner-visible documents.
Grader defect
A correct result fails an unstated ordering rule.
Evidence. The reference returns [a, b]. A grader expecting [b, a] rejects it even though the task does not require an order.
Decision. Correct the grader. Compare the expected IDs without imposing order, and separately check uniqueness and input preservation.
Specification gap
“Documents the user can access” is not a complete task.
Evidence. That shorter draft leaves tenant boundaries, archived records, ownership and duplicates undefined.
Decision. Clarify these rules before collecting judgments. Do not classify a solution as wrong based on requirements absent from its task.
Includes the fixture, both implementations, checks and limitations.
Read the complete authored sample04 / Engagement scope
Define the engagement around your work.
Discuss the engineering role, working arrangements and requirements directly with Quave—whether you represent a lab or its human-data partner.
Skills and availability
Which software domains, languages and experience does the work require? Discuss the people, capacity, schedule and rates for that scope.
Tasks and criteria
What are the expected outputs and acceptance criteria? Identify representative tasks and the questions that need the task owner’s judgment.
Workflow and review
Which tools and handoffs would the engineers use? Define review responsibilities, calibration and how disputed findings should be resolved.
Access and confidentiality
What restrictions apply to data, tools, retention and subcontracting? Discuss these boundaries before sharing project materials.
Contact
Tell us where you need engineering support.
Share your software domain, the type of work and the role you need to fill. We’ll discuss fit, availability and commercial terms.
contact@quave.ai