How Kepler Treats FDE as Product Strategy

These are my notes from the Forward Deployed Engineering workshop at @aiDotEngineer on June 30, 2026.

Session: How Forward Deployed Engineering Is Done at Kepler
Speaker: Vinoo Ganesh
Company: Kepler
X: @VinooGanesh

FDE as Product Strategy

Kepler's framing was direct: FDE is a product strategy, not a go-to-market or customer success function.

FDEs are judged by their ability to:

• Extend the product team.
• Identify opportunity areas.
• Generalize solutions.

Project Frontline trained about 350 software engineers as forward-deployed, product-focused individuals. Tally is building this function at Kepler using the same framework.

Detect the Real Problem

Customers describe solutions, not problems.

A shipping company's 40-page BI/dashboard requirement collapsed into a four-hour solution after on-site observation.

FDEs should understand:

• The core goal.
• What happens after the solution.
• How the customer solves it today.

Small, curtailed solutions build trust and keep the FDE as system owner.

Actions Speak Louder

A CSV-to-Parquet migration was blocked for a year because one user needed to double-click files on Windows for spot checks.

A Parquet viewer built overnight unlocked the migration. Pipeline time dropped from 17 hours to about 2 hours.

On-site signals include:

• Repeated tasks.
• Copy-paste between tools.
• Frustration.
• Tool switching.
• Phone use mid-workflow.

Ontology and Language

FDEs map organizational nouns and verbs.

Overloaded terms create:

• Broken integrations.
• Data quality issues.
• Failing pipelines.

Defining vocabulary inside the platform can make the product the organization's ontological foundation.

Ship Fast, But Build for Production

Every hack may become production.

Calibration questions:

• Will I get a 3 a.m. call about this in six months?
• Am I making the right tradeoff for the product?
• Who owns this when I leave?
• What happens when it breaks?

The default: treat every solution as if it will run for 18 months.

My Take

Kepler's framing is the most product-oriented: FDE is product strategy, not GTM or customer success.

The best examples were not about writing a lot of code. They were about seeing the small operational blocker that made the larger technical plan impossible. That is why on-site observation, vocabulary mapping, and ownership questions matter so much.