Software Engineer - Full Stack AI
Fullstack engineer who owns FraudX — a document fraud detection platform for Australian lenders — end to end: building a deterministic PDF inspection engine, PostgreSQL data models, and React/Node product surfaces on AWS (Lambda, ECS, S3), using AI tools like Bedrock and Claude Code in a test-first, written-spec workflow.
- Compensation
- A$100K–A$120K / yr
- Location
- Cremorne, Melbourne VIC
- Work setup
- Onsite
- Seniority
- Not specified
- Employment
- Not specified
- Experience
- 2+ years
- Category
- Fullstack
- Posted
Skills
Must have
- 3+ years of professional software engineering experience
- Minimum 2+ years of professional experience in the required or related role
- Australian work rights and must be based in Melbourne, Victoria
- PDF internals: cross-reference tables, incremental updates, object streams, fonts and content streams (ISO 32000), plus XMP and EXIF metadata
- Strong proficiency in Node.js and React.js with Tailwind CSS
- Hands-on experience with PostgreSQL (schema design, queries, migrations)
- Working experience with AWS services, particularly Lambda, API Gateway, RDS, S3, ECR, and ECS
- Familiarity with Supabase (auth, JWT, real-time, database, or storage)
- Experience with or willingness to adopt AI-assisted development tools (Claude Code, AWS Bedrock, AI Solutioning)
- Ability to work autonomously and take ownership of deliverables
- Comfortable using GitHub: making, reviewing and creating PRs
Nice to have
- Experience or exposure to financial services, mortgage broking, fintech, or compliance environments
- Has built something driven by rules or a specification: rules engine, interpreter, validator, schema-driven configuration system, parser or small language
- Digital forensics, document security, image forensics or anti-fraud work experience
- Basic understanding of AWS
- Understanding of API-driven platforms, automation tools, or document processing technologies
- Tertiary qualifications in Software Engineering (majoring in AI), IT, Data Analytics, or related disciplines
- Familiarity with infrastructure-as-code
Full description
Salary: $100,000 – $120,000 per year + bonus (package)
`About Docuscan Australia
Docuscan Australia is a leading provider of AI SaaS solutions including intelligent document management, workflow automation, and compliance solutions supporting the mortgage, finance, and broader financial services sectors. Our platforms help businesses streamline operations, improve data accuracy, and meet regulatory obligations with confidence. As we continue to scale, we are looking for a motivated Fullstack AI Software Engineer to embrace the opportunity to take real ownership and shape products from the ground up.
About the Role
FraudX is a document fraud detection platform used by Australian lenders and brokers. Applicants submit payslips, bank statements, tax and identity documents; FraudX tells the lender whether to trust them.
We are building the next generation of its detection engine, and we are hiring one engineer to build it alongside the person who designed it. You own the implementation end to end: the module that examines files at the byte level, the engine that runs a growing library of declarative checks, the data model that makes every result reproducible, and the product surfaces that put it in front of clients.
The requirements arrive written down, with acceptance criteria, fixtures and recorded decisions. Within them, how it is built is yours. If you have wanted to work somewhere that writes things down and means it, this is that place.
Australian work rights are required for this position and candidates MUST be based in Melbourne, Victoria to be considered for this position.
Candidates must have a minimum of 2+ years of professional experience in the required or related role.
Key Responsibilities
- Design, build, and maintain full stack features across our SaaS platforms including A deterministic detection engine, from the file up.
- Directly with the design authority, every day. Requirements, acceptance criteria and fixtures come to you in writing. Decisions are recorded. When you think one is wrong, you say so with a recommendation and get an answer before the code changes
- You own the implementation. The structure of the code, the tests, the migrations, the release: yours, end to end.
- Test-first, fixture-driven. Every result a check can return has a real document that produces it, and the fixtures are the acceptance criteria.
- Model and maintain relational databases in PostgreSQL (via RDS and Supabase)
- AI-assisted, human-reviewed. We build with AI coding tools from complete specifications, and review what they produce as carefully as hand-written code.