AI/ML Engineer

About Public Pulse

Public Pulse builds the communication and decision-making layer that modern parliamentary offices rely on. We handle parliamentary data at scale. Thousands of interactions per week flow through our platform, paired with drafting, approval workflows, CRM tracking, and real-time collaboration tools.

Our platform is already in use across parliamentary offices in Australia, Canada, and New Zealand. We're expanding internationally and growing fast.

About the Role

Our drafting engine - PulseAI - is the core of what makes Public Pulse different. When a constituent emails their MP about a planning issue, PulseAI searches the office's knowledge base of past correspondence, policy documents, and parliamentary records, then generates a citation-verified draft response with inline source attribution. The staff member reviews, edits, and approves before it goes out. Getting this right is not a feature - it is the product.

We are looking for an AI/ML Engineer to own and advance every layer of this system. You will work across the full pipeline: document ingestion and chunking, embedding generation, hybrid search (semantic + BM25 via pgvector), reranking, prompt construction, response generation with inline citations, and the evaluation frameworks that tell us whether any of it is actually working.

This is applied AI engineering in a high-stakes context. Government correspondence carries real consequences - a factual error in a constituent reply can become a media story. You will build the guardrails, evaluation systems, and human-in-the-loop workflows that make AI outputs trustworthy enough for parliamentary use.

You will not be training foundation models. You will be building the systems that make foundation models useful, reliable, and safe in a domain where accuracy and attribution are non-negotiable. You will work closely with our engineering team and directly with parliamentary staff who use the system every day - their feedback will be immediate and honest.

Responsibilities

• Own and optimise the full RAG pipeline - document processing, embedding generation, hybrid search, reranking, prompt construction, and response generation
• Build and maintain the knowledge base ingestion system that processes parliamentary correspondence, Hansard records, policy documents, and historical replies
• Design and implement citation verification and attribution systems - every AI-generated statement must trace back to source material
• Develop evaluation frameworks and automated testing for AI output quality, relevance, factual accuracy, and safety
• Integrate with foundation model providers (Anthropic Claude, Google Vertex AI) and optimise prompt engineering for government correspondence
• Build guardrails for AI safety - content filtering, confidence scoring, hallucination detection, and human-in-the-loop approval workflows
• Research and prototype improvements to retrieval quality, response accuracy, and system efficiency

You May Be a Good Fit If You

• Have 4+ years of experience in ML engineering, applied AI, or research engineering
• Have built RAG systems, embedding pipelines, or information retrieval systems that run in production - not just in notebooks
• Are proficient in Python and/or TypeScript and can build production ML systems with proper testing, monitoring, and deployment
• Have hands-on experience with vector databases, embedding models, and similarity search (pgvector, FAISS, Pinecone, or similar)
• Understand LLM integration deeply - prompt engineering, structured outputs, function calling, context window management, and the failure modes that come with each
• Know how to evaluate generative AI systems - you can design metrics, build evals, and tell the difference between a system that looks good and one that actually is good
• Communicate complex technical trade-offs clearly to non-technical stakeholders

Strong Candidates May Also Have

• Experience with responsible AI practices, content safety, or AI governance frameworks
• Background in information retrieval, NLP, or computational linguistics
• Prior experience in govtech, regulated industries, or high-stakes AI applications
• Experience with Anthropic Claude, Google Vertex AI, or similar provider APIs in production
• Published research or significant open-source contributions in retrieval, NLP, or applied ML

About Public Pulse

Public Pulse is building the operating system for parliamentary offices. We help government teams manage constituent correspondence with AI-powered drafting, approval workflows, and electorate intelligence - all with government-grade security. We are a small, focused team in Sydney building something that matters.

What we are looking for

• 4+ years ML engineering, applied AI, or research engineering experience
• Production experience building RAG systems, embedding pipelines, or information retrieval systems
• Strong Python and/or TypeScript with production ML deployment skills
• Hands-on with vector databases, embedding models, and similarity search (pgvector, FAISS, etc.)
• Deep LLM integration knowledge — prompt engineering, structured outputs, function calling, failure modes
• Experience designing evaluation frameworks for generative AI systems
• Clear communication of technical trade-offs to non-technical stakeholders
• Experience with responsible AI, content safety, or AI governance is highly valued
• Govtech, regulated industries, or high-stakes AI experience is a strong plus