I work on the infrastructure underneath AI features.
Small engineering teams ship AI features faster than they can staff for them. What follows is spend nobody owns and infrastructure nobody reviews. That gap is what I work on.
I'm Foad Talsi. I work alone, from Paris, on the infrastructure underneath AI features — mostly AWS and Terraform.
Right now that means audits: reading a live AWS account and coming back with what it costs, what to change, and what the change saves. The same problem shows up earlier, when the infrastructure is written rather than when it's running — so I also build tooling for that end of it, currently a GitHub Action that checks Terraform generated by AI models before it reaches an apply. It's all the same layer: the AWS underneath AI features, written faster than anyone has time to review it.
I'd rather be judged on output than on a CV, so the method is public. The reference environment listed above is a full AI stack on AWS, built wrong on purpose, with the complete audit report next to it — the same checks and the same format a client receives. Read it, then decide whether this is worth a conversation.