Forward Deployed Engineer

# Forward Deployed Engineer Apply via email
** Team:
** Solution Engineering ·
** Level:
** Mid-level, 3+ years## About distil labs

We are building a “distil-agent” that replaces LLMs with custom SLMs to automatically improve the efficiency of AI products. Customers generate savings of 50-80% and cut latency in a day – the platform automates the data curation, model fine-tuning, evaluation and deployment, so you can focus on shipping features. Under the hood we use knowledge distillation from large language models into lightweight student models that are as accurate as models 30-500x larger.
** Why now?
** The industry has converged on the view that small, specialized models are the workhorses of agentic AI: most production AI tasks are narrow, repetitive, and cost- and latency-sensitive – exactly where custom SLMs beat general-purpose LLMs. Enterprises are now moving these workloads from LLM prototypes into production, and they need a partner who gets their models adopted there.## About the role

You own everything that happens after a customer decides to work with us. Three things define the job:
1.
** Be the deepest power-user of the distil labs platform.
** You leverage the distil labs platform to create high-quality custom models that fulfill customer requirements, and you turn what you learn in the field into concrete product improvement requests for our engineering team.
2.
** Deliver models that clear the customer’s bar.
** You own the Solution Engineering process end to end, including the project management around every engagement. Success means two things: the model passes the customer’s own evaluations, and the customer trusts you enough to shift production traffic to distil labs endpoints. Your metric is time to model adopted in production.
3.
** Iterate on the developer/agent-facing interface to our platform (today, a Claude skill).
** Every engagement teaches us where the workflow is confusing or where a step should be automated. You fold those learnings back into the interface so every subsequent engagement runs faster and more reliably.

You will not build custom code for customers – you will build and integrate custom models for them. That is how a forward deployed engineer contributes at distil labs: your deliverable is a model in the customer’s product, not code in their infrastructure. Your leverage comes from operating the platform better than anyone else and running delivery like clockwork. You feed insights back to the product team to keep improving the platform.##

What you’ll doThe three pillars above define the job. On top of them, concretely, you will:
* Agree the evaluation criteria and acceptance bar with the customer up front, then deliver a model that clears it on their own evals, not just ours.
* Build the trust that makes customers confident to shift traffic: transparent progress updates, honest reads on model quality, and a controlled traffic ramp from first percent to full production.
* Run delivery like a project manager: set timelines, drive the daily cadence in shared customer Slack channels, surface blockers early, and push every engagement toward model adoption in production.
* Operate the distil labs platform at expert level: prepare data and job descriptions, configure trace processing, launch training, interpret evaluation metrics, and iterate the levers that move model quality.## What we’re looking for
* 3+ years in a technical, customer-facing role: forward deployed engineering, solutions engineering, ML or AI customer engineering, implementation, or similar.
* Hands-on fluency with LLMs in production: prompting, evaluation, and the failure modes that only show up fortable in the weeds of data: traces, JSON/JSONL, YAML configs, CLIs.
* Experience designing and…