Forward Deployed Engineer

TL;DR

We are seeking an entrepreneurial and driven individual with a keen interest in early‑stage startups and consulting experience in Data Science, particularly with projects involving Knowledge Graphs and Large Language Models (LLMs).

Why this role exists

Every large enterprise we talk to has the same problem: their GenAI pilots hallucinate on their own data, and nobody trusts the output enough to put it in production. We solve this by combining LLMs with knowledge graphs, and it works. Large Defense, Banks and Manufacturers are already in production with us.

Now we need someone to sit on the customer's side of the table and make sure every deployment becomes a reference. Not a Customer Success Manager with a technical vocabulary but an engineer who writes code in the customer's repo, designs the ontology with their domain experts, and ships the integration that makes graph‑based agents part of how the company actually works.

You’ll be one of the second Forward Deployed Engineer at Lettria and help write the playbook for this function.

Why Lettria

The big labs are hiring FDEs in Paris too. Here’s what’s different here:

• You’ll be employee ~20, not ~2,000. What you learn on a deployment changes the product the following sprint. You’ll talk to the founders daily.
• The technical bet is sharper. Generic RAG is a commodity. Agents grounded in structured knowledge is where regulated industries actually trust GenAI. You’ll go deep on graph modeling, ontology design, and graph‑native retrieval, not just prompt engineering.
• Customers are decision-makers, not pilots. Our buyers are Heads of Data, CDOs, and Chief Innovation Officers signing real contracts. You won’t run lighthouse demos that go nowhere.
• Trust is the product. Our entire wedge is making GenAI explainable and auditable enough for nuclear engineers or investment bankers. If you care about AI that is safe because of how it’s built, not because of guardrails layered on top, you’ll feel at home.

Tasks

What you’ll actually do

• Embed with customers. Spend real time on‑site with our enterprise accounts — understanding their data, their workflows, and the political map of who needs to be convinced. Co‑design the GraphRAG architecture with their engineering and domain teams.
• Build in their stack. Write production code that integrates Lettria into customer systems: connectors to their data sources (SharePoint, S3, internal APIs, document stores), ontology‑aware ingestion pipelines, custom retrieval logic, evaluation harnesses tied to their KPIs. Some of this is Python; some is graph query work (Cypher, SPARQL, GQL); some is plumbing across whatever the customer already runs.
• Own the deployment end‑to‑end. From kickoff to “this is in production and the business owner is referenceable.” Scoping, sequencing, escalating, unblocking.
• Turn deployments into a product. Every customer engagement should leave behind reusable assets — connectors, skill templates, eval datasets, ontology patterns. Codify them so the next deployment is faster than the last.
• Shape the playbook. As one of the founding FDEs, you’ll define how this function operates: scoping rituals, deployment checklists, what “done” looks like, how we measure success.
• Represent Lettria externally. Conferences, customer roundtables, technical content. We need credible voices in the field on knowledge graphs + LLMs.

Requirements

What we’re looking for

We care more about trajectory and judgment than ticking every box. If most of the below resonates, apply.

• 3+ years building and shipping software in customer‑facing or high‑ambiguity contexts, FDE, solutions engineering, tech consulting, early‑stage startup, or a strong engineering background paired with the willingness to sit across from a customer;
• Production experience with LLM systems. RAG pipelines, agents, evaluation, prompt engineering at a level beyond demos. You’ve debugged why a retrieval step is failing on real data, not just on the cookbook example;
• Strong Python. Comfortable owning a service end‑to‑end. Bonus for TypeScript/Node, Go, or Java if customer stacks demand it;
• Genuine interest in graphs. Hands‑on experience with Neo4j, Neptune or similar is a strong plus. If you haven’t worked with graphs yet but the idea of modeling a business domain as an ontology lights you up, that’s also a signal;
• You can sit with a CDO and a junior data engineer in the same meeting and add value to both conversations;
• Fluent French and English. Our customers are worldwide; the team and the codebase are in English;

Nice‑to‑haves

• Experience with semantic technologies (RDF, OWL, SHACL) or knowledge engineering;
• Background in a regulated vertical (financial services, public sector, healthcare, legal);
• Open‑source contributions to LLM, RAG, or graph tooling;
• You’ve been the first or second FDE/SE at a previous startup;

Process

Four conversations, one technical exercise, ~3 weeks end‑to‑end:

1. Intro call with the hiring manager
2. Technical deep‑dive with our Data Science / Engineering team (take‑home + discussion)
3. Customer‑scenario interview — a real (anonymized) scoping problem
4. Final round with the co‑founders

We give written feedback at every stage, whatever the outcome.