Forward Deployed Engineer

Conquer AI deploys production-ready AI systems for FTSE 100, FTSE 250 and Fortune 500 clients. This role sits on one of our engagements: a Fortune 500 healthcare and diagnostics client where our AI reads specimen paperwork and lab photos, pulls out the data a technician would pull out by hand, and sends anything uncertain to a person. Get this wrong and a real specimen gets mishandled, so the accuracy bar is high, and saying plainly where the model falls short matters just as much as the accuracy itself.

The reality of this role: You won't own one layer and hand off the rest. You will build the model pipeline, prove it works with real evaluation, build the interface the person on the other end actually uses, and run the AWS infrastructure underneath, usually inside a client's locked-down network. You'll also sit in front of the client: writing the validation reports, explaining a regression honestly instead of burying it, and adjusting the plan when the domain turns out messier than the deck suggested.

What you'll own

• Agent pipelines built on Claude via AWS Bedrock, using an agent framework such as Strands and MCP to connect the model to a client's internalAPIs
• Prompt engineering for document and image extraction, plus the reference data and confidence scoring that make it reliable rather than lucky
• Evaluation: ground truth datasets, accuracy scoring, and honest reporting when a model under performs
  
  The operator interface, a React or Next.js front end built for someone doing this job all day, including the human-in-the-loop screens for anything the model isn't sure about
• Deployment and infrastructure:
  Docker, Terraform and IAM, inside an enterprise network with its own registries, CI and security review
• Client-facing delivery: status updates, validation reports, and demos that show real progress rather than a polished slide

Who you are

• Fluent in Python and Type Script, and you've used an LLM API to build something that shipped, not a personal chatbot project
• You've worked with an agent framework and tool calling, ideally know MCP, and can prompt for structured extraction from documents or images
• You know how to tell if a model actually works: ground truth, scoring rules, denominators, not vibes
• Strong in React or Next.js, and comfortable owning a backend end to end
• Confident with AWS, Docker and Terraform, and can talk through an IAM setup without a diagram
• You've worked inside a client's constrained environment before, or you're ready to: internal registries, no public internet, Jenkins, SSOYou write clearly enough that a client could read your update without translation
• Genuinely curious about the domain you land in. We'll teach the healthcare and lab detail, but you need to want to learn it

Where you're coming from is a Software engineer or AI engineer who has shipped a real LLM system and measured it properly, and wants to work directly with a client rather than behind a product rodmap.

What we're offering

• Ownership of a live client engagement from build through to client sign-off, not a slice of one
• Direct exposure to the infrastructure behind a Fortune 500 healthcare deployment, with onsite time at kickoff, midpoint and leadout

If you'd rather be the engineer a client calls when the model gets something wrong, we should talk.