A former test engineer's first year as an FDE: business context and knowing who to ask beat knowing every tool

Manas Verma, an AI forward deployed engineer based in Pittsburgh, wrote a first-person piece for LeadDev about moving into the role earlier this year from a job as a software development engineer in test focused on data quality. His company formed an FDE team as customer demand for AI demos grew, and he says the skills that carried over were automation, knowledge of the ETL lifecycle, and explaining complex systems to non-technical people.

The technical list he gives is familiar: Python, SQL, REST APIs, Docker, and agent frameworks such as LangGraph and n8n. But he argues that understanding the client's business, gathering and validating requirements, and knowing which product expert, infrastructure team, or data owner to pull in matter just as much. One of his main projects is migrating a client's legacy data with no data dictionary, which meant profiling the data and mapping tables by hand rather than building agents.

He also describes using his company's approved AI assistant to search meeting notes and the internal knowledge base, while warning that it invented column names when mapping tables, so he reviews everything it produces. His view is that the role is shifting from AI demos toward reliable production systems built on a solid data layer.

For people trying to break into FDE work, the piece is a useful counterweight to agent-heavy job ads: his advice is to pick a real problem, talk to the people who live with it, break requirements into simple steps, surface edge cases early, and fit the technology to the problem rather than the other way around.