Real, hands-on applied AI with genuine client impact
Pros
Codex is a great place to do real, end-to-end applied AI and data science. In a short time, I got to build production-grade systems across very different domains - computer vision, document AI, LLM and RAG pipelines, and optimisation for a range of Australian businesses, all on AWS (Bedrock, SageMaker, Textract, Lambda). The work is genuinely impactful: projects delivered measurable results for clients, such as large reductions in processing time and cost, rather than staying as proofs of concept. There is a strong engineering culture around careful evaluation and keeping models grounded and reliable, which I really valued. You get a lot of ownership and autonomy, exposure to the full lifecycle from problem framing to deployment, and the chance to work directly with clients. Colleagues are smart, collaborative, and generous with their knowledge, so you learn very quickly. For anyone who wants to grow fast in applied AI and actually see their work used, it is an excellent environment.
Cons
As with most consulting, the pace is fast and you context-switch between clients and domains, which can be demanding at times. Being a lean team also means you wear several hats. For me, none of this outweighed how much I learned and enjoyed the work.
