Companies Are Retraining Their Own Staff Into Forward Deployed Engineers

Hiring cannot keep up, so the conversion starts inside

Indeed data shared with Business Insider put forward deployed engineer postings in April 2026 at 5,230% above their January 2025 level. That is an index against a January 2025 baseline, not a count of listings. The same Indeed data puts pay for the role at about $170,000 to over $200,000. Business Insider also noted that the consulting industry is leaning on forward deployed engineers.

Some large companies are now building the role out of people they already employ. KT will develop more than 500 members of its AX (AI transformation) division into FDEs over the next two years, Chosun Biz reported on August 17, citing industry sources. The report describes the job as reading a client’s business structure and open problems, then combining AI, data, and cloud technology into a working fix. It does not end at the proposal. System build, operations, and later upgrades belong to the same person. The 500 are not new hires. They already work in the division.

Japan’s big IT services firms are moving the same way. Nikkei xTECH reported on August 17 that the FDE strategies of Fujitsu, Hitachi, NEC, and NTT Data came into view at earnings briefings held from July to early August, with firms looking to both acquire and develop FDEs, and its headline framed the push as a move away from billing by the person-month. According to Nikkei, Hitachi said with its April to June results that it plans to reach 1,000 FDEs in Japan by the end of fiscal 2026.

Korea’s Software Policy & Research Institute (SPRi) spelled out the supply problem in a report on September 11. Citing Bank of Korea estimates, it put growth in Korea’s AI workforce at 5.3% a year from 2010 to 2024, while a JobKorea analysis of AI-related job postings showed a 112% rise over five years. New graduates alone cannot close that gap quickly, the report argues, so the fastest and most cost-effective route is to add AI transformation skills to people who already hold domain expertise. Tacit knowledge from the field builds up over a long time and is hard to teach quickly. The skills of using and coordinating AI can be learned in a relatively short stretch of intensive training and real application.

The diagnosis covers three areas: industry, technology, solutions

KT started by assessing division members on industry, technology, and solution capability. From those results it designs an individual career development path and links it to training, project work, and certification. Scenario exercises modeled on real business, on-the-job training, and expert mentoring are meant to take each person from defining a problem through building a minimum viable product to pitching it to a client. The reasoning given for the shift: as AI technology becomes broadly comparable, understanding an industry and producing real results matter more than simply owning a technology.

Which area is thin depends on where a person comes from. An engineer from a systems integration delivery team usually has the technology box filled but may never have followed a client’s workflow from start to finish. Someone from presales or consulting brings deep industry context and far less time spent building and shipping. SPRi’s list of core FDE skills asks for both in one person: defining and solving problems, data analysis, understanding and building with AI and software, industry and business knowledge, and communication.

Agent-based work adds another layer. SPRi says retraining should go past prompt basics into delegating work to AI agents, designing their permissions and policies, orchestrating several agents at once, and reviewing outputs while taking responsibility for them. For hands-on practice it proposes connecting internal data and systems to agents and turning work instructions into written specifications. Anyone who can explain the permission structure of the internal systems they use every day starts ahead here.

Some vendors already organize FDE teams by industry. Databricks has an open Senior Engagement Manager role inside its Forward Deployed Engineering organization serving Financial Services customers. It asks for 7 to 10 years running a professional services line of business and 10 or more years in senior customer-facing roles.

After the classroom comes a squad at the client site

The first stretch of a conversion happens in training, one full loop from problem definition to MVP. The next happens in the field. KT plans to deploy squads that put industry and business specialists together with AI, data, and cloud staff. At sites such as university hospitals, where data security makes it hard to take data outside, the squads would design solutions on real data and carry them through field validation and service development. A squad setup means nobody has to fill every box on day one. You join on your strongest area and build the weaker one alongside teammates. The step after that is owning a single client account, as a squad, all the way through.

The weak spot is proof. In a Wanted Lab survey cited by SPRi, companies hiring AI talent named accurate skill verification as their biggest difficulty, at 57.9%. SPRi proposes an AX skills framework that pairs assessment on real projects and deliverables with tiered credentials such as micro-credentials, but it remains a policy proposal. Until something like that exists, the evidence of a conversion is what you built: the MVP, the proposal you put in front of a client, and a record of whether it made it into production.