Forward Deployed Engineer, Research

Ersilia; San Francisco (CA); On-site; USD 140,000 - 190,000; Full time

Job summary

Ersilia is seeking a Forward Deployed Engineer to join a customer’s team and own outcomes end to end. You will embed with a specific customer, shaping how their agents and models perform in real usage.

You will design data pipelines from live production logs, test live models, and connect data changes to model behavior. You will work closely with researchers, engineers, and customers to deliver tangible results.

Qualifications

• Experience building AI agent systems used by real people or forward-deployed experience with direct coding ownership.
• Strong software fundamentals across a modern stack, backed by CS background and real software engineering.
• Ability to work with highly technical customers, researchers and engineers, not just business buyers.
• Comfort owning a customer relationship directly and being accountable for outcomes.

Responsibilities

• Turn live usage into training data by building pipelines from production logs, converting signals into training data without manual labeling.
• Test models in live deployment with scoring that scales to customer traffic and protects workflow quality.
• Trace data changes to model behavior to reduce guessing in the training loop.

Skills

Agent systems

Forward-deployed

Software fundamentals

Technical customer credibility

Customer ownership

Job description

About the company

The company works on the data and infrastructure problems behind frontier model training and deployed AI agents. It partners directly with labs on post-training research and infrastructure, and deploys models and agents for large enterprises and government customers.Less than four months into operation, the company is already profitable with over seven figures in realized revenue and additional contracted revenue. The small team includes engineers and early employees from leading enterprise software, AI, and semiconductor companies, supported by early institutional and angel investors including senior researchers from leading AI labs.

About the role

As a Forward Deployed Engineer, you sit inside a customer’s team and own the work of making their

agents and models better. You are not shipping a generic product. You are embedded with a specific

customer, working from real usage, and accountable for that relationship’s outcomes end to end.

What you will do

Turn live usage into training data: Build pipelines that pull signal from production logs, including which

answers users keep, override, or later correct, and turn that signal into training data without a human

labeling step in the middle.

Test models that are already live: Build scoring that is inexpensive enough to run on customer traffic,

tuned to what each customer considers good, and sensitive enough to catch a broken workflow before

overall metrics move.

Trace what data caused what: Build systems that connect changes in model behavior to the data that

drove those changes, reducing guesswork in the training loop.

What we are looking for

Agent systems or FDE experience: At least one year of significant experience building AI agent

systems used by real people, or technical forward-deployed experience with direct coding ownership.

Software engineering: Strong software fundamentals in any stack, supported by a computer science

background and real software engineering experience.

Technical customer credibility: Ability to work effectively with highly technical customers, especially

researchers and engineers rather than only business buyers.

Customer ownership: Comfort owning a customer relationship directly and being accountable for

outcomes.

Schedule and location: Able to work six days each week and be based in San Francisco. New York

candidates may be considered after spending their first 3-6 months in San Francisco.