Forward-Deployed AI Engineer: Build Prototypes Fast

Requirements

• 5+ years as a software engineer shipping working systems; recent hands-on experience building LLM-powered or agentic applications.
• Track record of embedding with users or customers — consulting, startup founding engineer, FDE, solutions engineering, internal platform teams, or similar.
• Comfortable owning a product from problem discovery through to working prototype and iteration, and then handing off cleanly to a production team.
• Strong full-stack or backend engineering fundamentals — Python, TypeScript/Node.js or similar — and the willingness to touch whatever layer the problem needs.
• Practical fluency with modern LLMs, agent frameworks, RAG, tool use, evaluation and observability tooling.
• Cloud-native experience on AWS and/or Azure (GCP a plus), with a working understanding of identity, data and deployment patterns.
• Daily use of AI-assisted developer tools and agentic coding workflows — you build with AI, not just about it.
• Able to read a business: you understand revenue, cost, workflow and decision-making, and you use that understanding to shape what gets built and why.
• Comfortable framing a problem as a business case, defending trade-offs and knowing when to stop building.
• Commercially aware — you understand that a useful prototype that people actually adopt beats an elegant one that nobody uses.
• Bias to action: you’d rather ship a rough v1 and learn than design the perfect v3 in a deck.
• Product sense: you care about the user, the workflow and the outcome — not just the code.
• Comfortable with ambiguity — briefs change, data is incomplete, priorities move; you keep shipping.
• Teacher instinct: you make the people around you better without being asked to.
• (Desirable) Founding or early engineer experience at a startup.
• (Desirable) Experience in regulated industries (financial services, insurance, healthcare, energy, public sector).
• (Desirable) Open-source contributions or public writing/speaking on agentic systems.

What the job involves

• Forward Deployed Engineers (FDEs) are the people we put in the room when a problem is ambiguous, the data is messy, and a working solution needs to exist in weeks, not quarters.
• As an FDE at Elixirr, you will embed with our delivery teams and internal functions to go from idea to working prototype, and to build the internal tools and agents that accelerate how Elixirr itself delivers.
• We’ve deliberately modelled this role on how the category is defined in the market — pioneered at Palantir and scaled by firms like OpenAI.
• An FDE is a hybrid of technical lead, consultant and product manager: someone who writes production-quality code, but is just as comfortable sitting with users, understanding how decisions actually get made, and stress-testing a prototype in the field.
• Business acumen is not a bonus here — it is core to the job.
• An important scope point: this role is focused on idea-to-prototype and on building internal solutions that make Elixirr faster and better.
• A separate, dedicated team owns productionization into enterprise-grade software.
• You will work closely with them — and hand over cleanly — but the core of your job is learning fast, building fast, and proving value.
• Deploy into Elixirr delivery teams and internal functions to identify the highest-leverage AI opportunities inside real workflows.
• Sit with consultants, operators and end users to understand how decisions are actually made — then design around what you see, not what was described in a deck.
• Translate ambiguous business problems into concrete agentic solutions with clear success metrics and an honest view of the business case.
• Take ideas from napkin to working prototype in days, not quarters.
• Design, build and demo agents and agentic applications that execute real work — document processing, research, analysis, operational automations, internal copilots, workflow accelerators.
• Use modern agent frameworks and tooling (LangGraph, Semantic Kernel, AutoGen, CrewAI, MCP, A2A) pragmatically — pick the simplest thing that works.
• Instrument prototypes properly from day one: traces, evaluations, cost, latency and user feedback so decisions are grounded in data, not demos.
• Ship internal agents and tools that measurably accelerate how Elixirr delivers — from research and analysis to proposal shaping, document production and operational tasks.
• Extract reusable components, patterns and accelerators from each engagement back into Elixirr’s internal platform so the next team starts further up the curve.
• Run tight feedback loops with the consultants and operators using what you build, and iterate aggressively on what’s actually useful.
• Partner with Elixirr’s production engineering team when a prototype earns the right to become enterprise-grade software.
• Package prototypes for handover: clear documentation, evals, architectural notes and known limitations — so the production team can industrialize quickly and safely.
• Stay engaged as a subject matter expert during productionization without becoming the bottleneck.
• Understand the economics of what you build — where value is created, where cost sits, and what a realistic adoption path looks like inside a real organization.
• Navigate security, compliance, change management and data realities without using them as reasons to slow down.
• Be the trusted technical voice leaders turn to when they want to know whether something is real or a demo — and say so honestly, either way.
• Where engagements are client-facing, engage credibly with senior client stakeholders, frame trade-offs clearly and keep the conversation grounded in outcomes.