Solutions Engineering Is Reinventing Itself as Forward Deployed AI Engineering

TL;DR#

The Solutions Engineer role is being absorbed and re-expanded as Forward Deployed AI Engineering — the biggest org-chart shift in enterprise software since DevOps emerged in the late 2000s. Across 30+ SaaS companies tracked through 2025-2026, traditional SE job postings have dropped 14% while Forward Deployed Engineer (FDE) and Applied AI Engineer postings have grown 380% year-over-year. Anthropic, Palantir, Harvey, Cohere, and Mistral have publicly restructured customer-facing technical teams around FDE pods, and Databricks, Scale AI, and Snowflake have followed with internal renames. Comp data from 1,200+ practitioners shows the SE-to-FDE transition delivers a +35% average total-comp lift in 2026 levels, with frontier-lab FDEs clearing $400K-$650K all-in. The shift is structural: AI-first products need engineers embedded in the customer's workflow, not slide-driven demo specialists. For the 80,000+ SEs currently in market, the transition window is real but narrow — the differentiating skill is shipping production code against a customer's data, not running discovery decks. This post traces the data, the absorbed responsibilities, the comp trajectory, and what it means for individual practitioners.

What is the SE-to-FDE transition?#

The SE-to-FDE transition is the structural conversion of presales Solutions Engineer roles into hybrid engineering-plus-customer roles where the practitioner ships production code, owns deployment outcomes, and operates as a quasi-member of the customer's engineering team. Where the SE optimized for demo polish, technical qualification, and RFP responses, the FDE optimizes for prototype velocity, deployment depth, and outcome ownership — usually on top of an LLM platform or applied-AI product. The role originated at Palantir's forward deployed engineering function and was copied wholesale by Anthropic, OpenAI, and Cohere starting in 2023.

This isn't a rename. The job is materially different on three axes — code output, deployment depth, and revenue accountability. Our 2026 state of FDE survey of 1,500 practitioners found that 71% of FDEs ship merged pull requests into customer repositories monthly. The number for traditional SEs in 2024 was under 5%.

The data: SE postings down 14%, FDE postings up 380%#

Job board scrapes from LinkedIn, Wellfound, Y Combinator's WorkAtAStartup, and Greenhouse-powered career pages show the migration in stark numbers. Between Q1 2024 and Q1 2026:

• "Solutions Engineer" postings (SaaS, US): down 14%, with the steepest drops at infrastructure and developer-tools companies (-22% and -19% respectively).
• "Forward Deployed Engineer" postings: up 380%, a base-effect number that nonetheless reflects real volume — over 4,100 distinct openings posted in 2025 alone.
• "Applied AI Engineer" postings: up 240%, with significant overlap to FDE in scope and seniority bands.
• "Solutions Architect" postings: up 7% — the title is surviving better than SE, but increasingly with AI/ML qualifications appended.

The data lines up with what hiring managers say in our customer discovery interviews with founders building FDE functions: the bar moved from "can explain the product" to "can ship something the customer would actually deploy by end of week."

Cross-referenced against Bureau of Labor Statistics OEWS data on software engineering occupations, the FDE category sits in the upper-quartile compensation band — closer to ML Engineer than to Sales Engineer (SOC 41-9031), which is the bucket SE has historically reported into.

What FDE absorbs from the SE role (and what it adds)#

The FDE role inherits the customer-facing posture of an SE but adds production engineering responsibility and outcome accountability. Five concrete handoffs define the absorption:

1. Discovery calls → embedded discovery sessions. FDEs run customer discovery the way FDE-driven startups outpace sales-led AI competitors: in the customer's repo, on the customer's data, often live-pairing with a customer engineer.
2. Demos → working prototypes. Where an SE built a sandbox demo, the FDE checks code into a feature branch on day one.
3. RFP responses → deployment plans. The artifact shifted from a 40-page security questionnaire to a deployment doc with concrete latency budgets, model selection rationale, and eval criteria.
4. POCs → production pilots. FDE pilots typically run 4-8 weeks against real customer data and end with a measurable outcome, not a "successful POC" green-light.
5. Handoff to CS → no handoff. FDEs own the customer through expansion. CSMs handle commercial relationship and renewals; FDEs handle technical depth.

The function adds two responsibilities SE never had: writing and shipping evals (LLM-product correctness suites), and contributing back to the platform team (FDEs are typically the primary internal channel for "what's broken with the product when it hits a real customer").

For a tool-level view of what FDEs work in day-to-day, see the comparison of the best tools for forward deployed engineers in 2026 and the FDE tech stack breakdown.

Comp trajectory: SE → FDE = +35% on average in 2026 levels#

The single most important number in the transition is comp. Drawing on our 2026 forward deployed engineering compensation report covering 1,200 FDEs, the average lift on the SE-to-FDE move is +35% on total comp, with the breakdown:

• Mid-level SE ($165K-$200K total) → Mid-level FDE ($220K-$285K total): roughly +32%.
• Senior SE ($210K-$260K total) → Senior FDE ($310K-$420K total): roughly +44%.
• Principal/Staff SE ($270K-$320K total) → Staff FDE ($430K-$650K total): roughly +51%.

Frontier-lab FDE bands (Anthropic, OpenAI, Cohere) clear $400K all-in at the senior level and run past $650K at staff with equity. The compensation premium prices in both the engineering depth (these are bona fide IC SWE roles) and the revenue accountability (FDEs often have OTE-style variable comp tied to deployment milestones).

This isn't an arbitrage that lasts forever. Comp premiums of this size historically compress within 18-24 months as supply catches up. The differentiating skill that holds the premium is shipping in customer environments — exactly the muscle SEs do not develop in their traditional role.

Companies leading the transition#

Five companies are setting the pattern that the rest of the industry is now copying. The roster matters because it tells you which playbooks to study:

• Anthropic's Applied AI Engineering team runs an explicit FDE model where engineers spend 50%+ of time in customer environments. The team's hiring bar is closer to senior SWE at FAANG than to traditional presales — see the Anthropic Applied AI Engineer interview process breakdown for what they actually test.
• Palantir's FDE function is the genealogical origin of the role and the template most labs copy. The Palantir playbook treats FDEs as the front line of product discovery, with explicit feedback loops into platform engineering.
• Harvey AI runs FDE pods embedded inside BigLaw firms — a model that treats deployment as a multi-month engineering engagement, not a presales motion.
• Cohere's forward-deployed strategy publicly positions enterprise LLM work as "build with customers," with FDE-led engagements at every named-enterprise reference customer.
• Mistral's FDE function is the European analog, with deep-pocketed government and regulated-industry deployments running through FDE-led pods.

Outside the labs, Scale AI is using FDEs as the wedge for RL data and enterprise annotation, and Databricks, Snowflake, and Datadog have begun renaming portions of their solutions organizations to "Field Engineering" or "Applied" titles with FDE-style scopes.

The pattern recursion is now hitting Series A AI startups — see why every Series A AI startup needs an FDE in the first 10 hires.

What this means for the 80,000+ SEs in the market today#

For practicing Solutions Engineers, the implication is direct: the role is bifurcating. The half of the SE population that develops production engineering depth — shipping merged PRs, writing evals, owning deployment outcomes — is migrating into FDE bands and capturing the +35% comp lift. The other half is competing for a shrinking pool of traditional SE roles at companies that haven't yet restructured.

Three concrete moves for the SE planning the transition:

1. Ship code in customer environments. Not internal sandboxes. Find any reason to write merged PRs into customer repos this quarter.
2. Build LLM-product eval skill. Eval design is the single most distinguishing skill between traditional SE and FDE work. Frontier labs interview heavily on it.
3. Pick a vertical depth. FDE work is converging on vertical depth: legal, finance, defense, healthcare, biotech. Generalist FDEs exist but the premium is at the vertical edges.

For hiring managers at SaaS companies, the implication is that traditional SE org charts are now actively eroding margin and pipeline velocity at the high end of the market. Customers buying AI-first products want technical depth they can lean on through deployment, not a slide-driven qualification motion.

For research and product teams, the FDE shift creates a structural advantage in customer learning. FDEs in customer environments generate more, deeper, faster signal than any survey or quarterly business review could. Forward deployed customer discovery is replacing traditional discovery surveys, which connects directly to the broader 2026 customer research methodology shift that this blog covers across the rest of the cluster.

Where Perspective AI fits#

Perspective AI sits adjacent to this shift. We're built on the premise that AI-first customer research cannot start with a web form — the same premise that makes FDE work effective. Where FDEs generate deep signal by being embedded in customer environments, Perspective AI generates depth by running AI-moderated interviews that follow up on vague answers in the customer's own words. The complement is direct: FDEs handle the engineering depth, Perspective AI handles the qualitative discovery layer for the rest of the customer base that doesn't get an embedded engineer. Teams running this combination are publicly discussed in our writeup on how Anthropic, Scale AI, and other frontier labs run customer research.

Frequently Asked Questions#

Is the SE role dying or just renaming?#

The Solutions Engineer role is not dying outright — it's bifurcating. Roughly half of SE roles are converting into FDE/Applied AI Engineer titles with materially expanded scope (production code, deployment ownership, evals) and +35% comp. The other half remain as traditional SE roles concentrated in mature SaaS categories with weaker AI exposure. The net headcount move across SaaS is -14% on SE postings and +380% on FDE/Applied AI postings between Q1 2024 and Q1 2026. The role isn't dead; it's being re-expanded.

Can a traditional SE transition into an FDE role without an engineering background?#

Yes, but the transition requires real engineering muscle, not just SQL and notebook skill. FDE interview loops at frontier labs (Anthropic, OpenAI, Cohere) include a production coding round and an LLM eval design round that don't appear in traditional SE interviews. The fastest transition path is: ship merged PRs in any customer environment this quarter, build one production-grade eval suite, and contribute to one open-source agent or LLM tooling project. Most successful transitions take 6-12 months of deliberate skill building.

What's the salary difference between an SE and an FDE in 2026?#

The average total-comp lift from SE to FDE in 2026 is +35%. At mid-level the move is roughly +32% (from $165K-$200K to $220K-$285K). At senior level it's +44% (from $210K-$260K to $310K-$420K). At staff/principal the move is +51%, with frontier-lab FDEs clearing $400K-$650K all-in. The premium reflects both the engineering depth required and the direct revenue accountability FDEs carry through deployment milestones.

Which companies are hiring the most FDEs in 2026?#

The largest FDE hiring volume in 2026 is at Anthropic, OpenAI, Palantir, Scale AI, and Databricks, followed by vertical-AI leaders Harvey (legal), Glean (enterprise search), and Cohere (enterprise LLM). Mistral leads European hiring. Outside the labs, Snowflake, Datadog, and ServiceNow have publicly restructured portions of their solutions organizations toward FDE-style scopes. Series A AI startups are now the fastest-growing source of FDE openings, with most posting their first FDE role between seed and Series B.

Is "Forward Deployed Engineer" just a Palantir term other companies copied?#

The title originated at Palantir but is now an industry-standard role description used by 30+ AI-first companies. Anthropic, OpenAI, Cohere, Mistral, Harvey, and Scale AI all use FDE or Applied AI Engineer titles with materially identical scope: embedded customer engineering, deployment ownership, eval design, and platform feedback. The pattern recursion through Series A AI startups has cemented FDE as a category, not a single-company term.

How does the FDE shift affect Customer Success and Account Management roles?#

The FDE shift removes the traditional handoff from presales (SE) to post-sales (CSM). FDEs typically own customer technical depth through expansion, while CSMs retain commercial relationship, renewals, and adoption metrics. The cleanest division in 2026 is: FDE owns technical outcomes and platform feedback; CSM owns commercial health and account expansion strategy. Companies that haven't separated these scopes tend to either burn out FDEs with renewal mechanics or lose technical depth at the customer.

The bottom line#

The SE-to-FDE transition is the most significant org-chart shift in enterprise software since DevOps emerged. It's not a title rename — it's a structural rebuild of how technical customer-facing work happens at AI-first companies. The data shows the migration is well underway: -14% on traditional SE postings, +380% on FDE postings, +35% comp lift on the role move, and 30+ companies leading the restructure. For practicing Solutions Engineers, the window to make the transition is real but narrow. The differentiating skill is production engineering depth in customer environments — and the labs are already pricing it accordingly.