Forward Deployed Engineer, Applied ML
The applied ML version of the FDE role. The model layer is the part you own, from adaptation and evaluation through production traffic.
Start with this role, then explore whatβs next in your career with Harper.
Company Description
A global AI company that reached a $5B valuation in under 2 years, opening its Korea office after New York, London, Singapore and Tokyo. The product is an enterprise AI operating system: agents that run inside a company's actual operations rather than sitting beside them.
This is the version of the role where the model is your responsibility, not a dependency you call. If you have been doing ML but have never sat in front of the customer whose problem it solves, this is the door into that.
This role is live and confirmed with the company that is hiring. Harper never posts a fake or placeholder job.
What You'll Do
Forward Deployed Engineering is not traditional product engineering. You sit with the customer, learn how their operations actually run, and build the systems that make AI work inside them. On this track, the model layer is the part you own.
- Embed with customers to understand their data, their workflows, and what working actually means to them.
- Adapt and evaluate models against the problem in front of you rather than a benchmark.
- Build the retrieval, evaluation, and observability that tells you whether it is actually working in production.
- Own the data: what gets collected, how it gets cleaned, and what that does to model behavior.
- Take a model from a notebook to something that serves real customer traffic, and keep it healthy there.
- Feed what you learn in the field back to the product and engineering teams.
Required Qualifications
- 2+ years as a technical individual contributor, with machine learning work that reached production rather than staying in research or coursework.
- Strong Python and PyTorch, or an equivalent framework you have gone deep in.
- Evidence of taking an LLM or ML application to production: fine-tuning, retrieval, or agent work applied to a real problem.
- You have measured a model's behavior with something you designed yourself, not only a standard metric.
- Willingness to be in front of customers. This is customer-facing engineering.
- Business-level Korean and English. You will work with the global team directly, with no layer in between.
Years of experience is not a hard cutoff. This company is hiring across levels in Korea at the same time. Talk to Harper and it tells you which level fits your track record before you talk to anyone.
Nice to Have
- LLM fine-tuning, advanced RAG, or agentic use cases running in production.
- Experience integrating models through APIs into backend or front-end flows.
- Distributed training, or work at a scale where infrastructure was part of the problem.
- Publications, open-source contributions, or models other people use.
- Cloud fluency (AWS, GCP, or Azure).
Why This Role Is Remarkable
- Korea has strong ML talent and very few seats where the model ships into a paying customer's operations. This is one.
- Forward Deployed Engineering is the role frontier AI companies are competing hardest to fill, and very few engineers in Korea have done it yet.
- The Korea team is early. What you build in the first year sets the technical direction here.