AI Platform · Full-time

AI Deployment Manager

Australia (remote-friendly) / hybrid Sydney preferred

Take the Vecta AI agent from prototype to reliable production: evals, latency, cost, safety and release discipline.

About Vecta

Vecta is building the financial command centre for people who take their money seriously. We connect banks, brokers, superannuation and cards into one live view, then layer a personal AI agent on top so customers can ask questions, spot anomalies and understand their full picture without hopping between apps.

We are an Australian company (O&C Technology Pty Ltd, trading as VECTA AI) building for CDR Open Banking first, with global expansion planned. The product is read-only by design. We never move money. Security, consent and clear explanations are part of the product, not afterthoughts.

You will join an early team with high ownership. We ship carefully in a regulated space, prefer evidence over slogans, and care about latency, reliability and trust as much as features.

About the role

You will be accountable for getting our proprietary Vecta AI agent into production and keeping it there. That means model and tool selection, retrieval over customer financial data, evaluation harnesses, guardrails, cost/latency budgets and rollout playbooks. You work closely with engineering and product, and you are comfortable being the escalation point when quality, safety or performance slips.

Key responsibilities

Own production readiness for the Vecta AI agent: reliability, latency, cost and quality
Design evaluation suites for factual accuracy on balances, transactions and holdings
Build deployment patterns for prompting, tools/RAG, versioning and canary releases
Define guardrails so the agent stays read-only, citeable and honest about uncertainty
Instrument live traffic: failure modes, regressions, token spend and user outcomes
Partner with engineering on serving architecture and with product on capability boundaries
Create runbooks for incidents, model changes and provider outages
Turn field issues into clear requirements for the roadmap

Qualifications

5+ years in software engineering, ML engineering, MLOps or applied AI systems
2+ years shipping LLM or ML features into production (not demos only)
Practical experience with evaluation frameworks, regression testing and release gates
Comfortable trading off model quality, latency, cost and safety with clear metrics
Ability to debug production AI issues across prompts, tools, data freshness and infra
Strong collaboration with product and engineering in a regulated or high-trust domain
Based in Australia or able to work overlapping Australian business hours

Nice to have

Experience with agent tooling, RAG over structured data or streaming responses
Familiarity with OpenAI, Anthropic or similar providers in production
Background in fintech, compliance-sensitive products or data privacy controls