Website:
mvp1.com.au
Job details:
About MVP Ventures
We build and ship AI-powered software products for Australian healthcare, aged care, NDIS, and enterprise businesses. Our systems are live in production — AI agents, LLM pipelines, voice automation, and sovereign private AI. We are a lean, fast-moving team that ships real things to real customers.
The Role in One Line
You will design and build AI-powered solutions across multiple industries — translating real business problems into working agentic systems, automations, and intelligent products.
What You'll Actually Work On
Most of your time is Python — async services, queue workers, API integrations, and careful debugging. You'll build LLM pipelines that produce structured, validated output, RAG retrieval systems, and agentic workflows. You'll also build and integrate Next.js frontend interfaces that bring those AI systems to life for end users — streaming responses, real-time agent state, and AI-powered product experiences. Each engagement spans healthcare, aged care, education, or enterprise — each with its own compliance requirements, data constraints, and user needs. You adapt. You ship.
Responsibilities
- Design and build AI agent workflows and automation systems for client businesses across multiple industries
- Build and integrate Next.js frontend interfaces for AI products — streaming LLM responses, real-time agent state, and interactive AI-powered experiences
- Prepare and curate domain-specific datasets across healthcare, education, enterprise, and other verticals
- Explore instruction tuning, domain adaptation, and prompt optimisation techniques
- Build and test RAG pipelines with vector databases (FAISS, Pinecone, pgvector, Weaviate) — integrate vetted knowledge sources and evaluate grounding quality and faithfulness
- Develop async Python services, FastAPI endpoints, queue workers, and AWS integrations (S3, SQS, Bedrock, Lambda)
- Implement prompt engineering, hallucination mitigation, LLM output validation, and safety evaluation layers
- Work with LLM evaluation benchmarks — hallucination, grounding, and factuality
- Contribute to sovereign AI deployments inside client cloud environments (AWS, Azure, GCP)
- Translate business problems into AI/ML solutions and communicate them clearly to non-engineers
- Become an integral part of the product, AI solutions, and AI agents teams — prototype and validate AI use cases across the business
- Collaborate across multiple industries and adapt models and systems to new domains
Qualifications
Must Have
- 3+ years building AI agent systems and automations for real industry projects
- Solid Python — async/await, FastAPI, raw PostgreSQL (no ORM), Redis, Docker, Git
- Proficient with PyTorch, Hugging Face Transformers, and LangChain
- You've used an LLM API for something real (OpenAI, Anthropic, Bedrock, OpenRouter — any of them). You understand prompts, tokens, and why a model's output isn't guaranteed
- Next.js — comfortable building and integrating frontend interfaces with AI backends, including streaming and async UI patterns
- Experience with vector databases and RAG pipelines
- AWS: S3, SQS, Bedrock, Lambda
- Experience with MCP servers, Kafka, n8n agent workflows, and CRM automations
- Good knowledge of SQL and NoSQL — we use PostgreSQL with raw queries, no ORM
- Git, Docker, and the willingness to read a stack trace to the bottom rather than guess
Nice to Have
- Knowledge of data compliance and security practices across industries
- Familiarity with multi-modal AI (vision + text, tabular + text, etc.)
- Experience with LLM evaluation benchmarks — hallucination, grounding, factuality
- Strong ability to adapt AI models across new domains and industries
What matters more than your tech stack
- You understand why a system is shaped the way it is — given a design, you can explain what breaks if a piece moves
- You debug before you assume — check what's actually running before theorising about why it's broken
- You ask before changing direction — if the approach seems wrong, say so and explain why. Don't quietly rebuild it your way
- You explain in plain English to non-engineers — "the form saved to the wrong folder" beats a paragraph of architecture vocabulary
- You care about correctness over cleverness — these systems handle sensitive data for real people. A missed field costs someone hours of manual work
How We Work
Small team. Direct access to the ML Engineer and CTO Advisor — no layers between you and the decisions. Architecture decisions are made deliberately and then locked; you extend well-defined systems, not rewrite them. Room to propose new things once you know the ground. Australian data residency is a hard requirement on most products — you'll learn to design within real compliance limits. Everything ships. If you build it, it goes to customers.
What You Get
- Production experience with LLM systems — agent orchestration, retrieval, voice, evaluation — at a depth most roles never reach
- Mentorship from an ML Engineer who is hands-on in the code daily, not managing from a distance
- Ownership of real components within your first month
- Collaboration with founders, engineers, and AI researchers
- Flexible remote environment
- Competitive salary with performance-linked component
- Active investment in your continued learning and growth into an AI Lead role
This is a remote India-based role. You will work across multiple industries and AI product lines simultaneously — adapting, building, and shipping AI solutions for real businesses. The engineer who grows in this role is our first candidate for AI Lead as the practice scales.
Click on Apply to know more.