Serverless AI Agent Backend Developer for a Multi-Tenant Cloud Build
The client wants a serverless AI agent backend developer to implement endpoints for a multi-tenant agent product from written specifications. The brief covers task-by-task delivery, validated slices and the serverless stack already chosen.
- Open ยท checked 12 Nov 2026
- ๐ฆ๐บ Australia (budget currency)
- Amazon Web Services
Is the budget fair? The budget looks tight for a multi-tenant backend with authorisation, queues, storage and monitoring. It is more workable if the first milestone covers a small number of tested endpoints.
What the client needs
- Implement tasks in the order set by the specifications.
- Build serverless endpoints in Node.js and TypeScript.
- Validate each slice with tests before moving on.
- Work from written requirements, design and task lists.
Skills the client asked for: Amazon Web Services, Node.js, Git, Artificial Intelligence, Typescript, AWS Lambda, NoSQL, DevOps.
Where to aim your bid
The sweet spot sits about 40% of the way up the client's range, the same rule AutoBidPilot uses for its own price suggestions. Go higher when the scope is large or unclear; go lower only when you can show close, finished work. Check what you keep after fees with the bid price calculator.
What the project involves
The client is building a multi-tenant AI agent platform for a SaaS ERP product. Users chat with agents that find and change data safely, with a preview and explicit approval before anything is written. It runs fully on AWS serverless and uses Amazon Bedrock for the models.
Specs already exist in Kiro format, with requirements, design and task lists per phase. The developer implements the task lists in order, validates each slice and delivers tested endpoints.
Stack
- AWS Lambda with Node.js and TypeScript, Serverless Framework v4
- API Gateway REST with WAF and a custom Lambda authoriser
- DynamoDB, SQS, S3, EventBridge, Secrets Manager and CloudWatch
- Amazon Bedrock
- GitHub with CI/CD pipelines
Required experience
- Real Kiro experience using spec workflow and steering files, validating output through tests, with examples
- Production Lambda and serverless work: event-driven design, SQS, idempotent workers, least-privilege IAM, cost awareness
- Git and CI/CD, such as GitHub Actions with OIDC across environments
- Strong TypeScript, automated testing and clean code
- Clear written English and regular progress updates
- Nice to have: Bedrock or LLM tool-calling and multi-tenant SaaS security
How the work runs and applying
A digital agreement on confidentiality and IP comes first. Delivery is in milestones with gates, and payment is released after testing and code review. Finishing on time or early earns an AU$250 bonus. Work happens in a provided dev environment without production access. Applicants without verifiable Kiro and AWS Lambda experience will not be considered.
Applicants should describe how they used Kiro and validated output, a shipped serverless AWS project with queues, testing and CI/CD, how they test an endpoint before calling it done, links to repositories or demos, and their availability and start date.
How to win this project
- Ask for the written specifications and task lists, so you can confirm how much of the design is already settled.
- Show a serverless project with authorisation, queues and infrastructure as code, and explain how you tested it.
- Propose one validated slice per milestone, which matches the milestone-based payment the client describes.
Questions to ask the client
A specific question in your first message shows you read the brief. Pick one or two:
- Which specification version and task list should the first milestone follow?
- Who sets up cloud accounts and access permissions?
- How will AI model usage be tested and monitored?
Common questions
How much does a serverless AI agent backend developer cost?
The listed budget is A$750โ1,500 AUD. Check how many tasks and slices a milestone covers before agreeing a price per milestone.
What should a serverless AI agent backend developer proposal include?
Examples of serverless endpoints you have shipped, your test approach per slice, the milestone plan and your cloud security setup.
Is this project still open?
We checked this project on 10 November 2026. Projects get awarded, so apply soon; we take a brief down when it closes.
This page is our own summary of a project posted online, written to help you decide and to write a better bid. Projects get awarded or closed, so apply soon; we take a brief down when it closes. No job or income is guaranteed.