Comp and Where FDEs Go Next

Verified Palantir comp from Levels.fyi, a frank acknowledgment of what non-Palantir comp data is reliable (very little), and three named ex-FDE founders with real funding totals.

Palantir FDSE comp, with the actual numbers

Per Levels.fyi (last updated June 2026), Palantir's Forward Deployed Software Engineer role has a median total compensation of $215,000 and a range of $171,000 to $415,000+. At the Software Engineer tier specifically, the breakdown is roughly $154,000 base / $60,200 stock per year / $12,800 bonus, which is $227,000 total. These are the only FDE comp numbers our second research pass could verify against a primary source. They are the reference point you can quote; everything else in this lesson is held to a lower confidence bar.

Non-Palantir comp: what we deliberately won't tell you

Multiple widely-shared comp figures for OpenAI, Anthropic, Scale, and Ramp FDE roles failed adversarial verification in our second research pass. Specifically: Perspective AI's '2026 FDE compensation report' (claiming a 1,200-FDE sample size, OpenAI L5 ~$1.0M, L6 ~$1.28M) was refuted on methodology; Hashnode's guide (claiming Palantir avg $238K, OpenAI/Anthropic $350K-$550K mid-to-senior) was refuted; Secondtalent's OpenAI base bands ($162K-$280K) were refuted. We are not going to repeat those numbers as 'the OpenAI FDE range' because we couldn't verify them. What we can say with high confidence from Anthropic's own posted job listing on Greenhouse: the 'Forward Deployed Engineer, Applied AI' role has a base of $200,000-$300,000 USD. Anthropic also has a public no-negotiation policy. Dario Amodei to Fortune (August 2025): 'We don't negotiate that level because we think it's unfair.' Total comp at Anthropic is higher than that base by a meaningful equity multiple, but specifically how much higher we will not invent.

Where ex-FDEs go: founding companies is real, but the denominator isn't

The Palantir FDE alumni network has produced a meaningful founder pipeline. Confirmed examples from our research: Gintautas Miliauskas, ex-Palantir FDE, founded Mavenoid, which has raised approximately $40 million total funding. Angela McNeal, ex-Palantir FDE, founded Thread AI, which has raised $26 million ($6M seed from Index Ventures in October 2024, $20M Series A from Greycroft in June 2025, the latter covered by Fortune). Engin Ural and Pratap Ranade, both ex-Palantir FDEs, co-founded Arena along with ex-Deployment Strategist Rob Speare; Arena raised $32 million from Initialized. The conservative 'FDE-specific subset' commonly cited is 40+ ex-FDE-founded companies, though related aggregate numbers (the often-repeated 111 Palantir-mafia companies, $11.6B raised, $150M raised specifically by ex-FDEs) all failed verification and are not used here. The honest framing is: founding-a-company is a real exit path with multiple named, funded examples, but we don't have a verifiable rate (founders per ex-FDE) and we don't know what the most common non-founder paths look like (PM, EM, staying IC at scale, returning to consulting).

The case for the role from a comp angle

If you're choosing between an FDE role and a regular SWE role at the same company, the headline numbers tell only part of the story. Two structural things matter. First, FDE roles at AI labs require production LLM experience and customer-facing fluency, which compress how many people can credibly apply, which historically pushes comp up at the senior end. Second, FDE work tends to put you in close contact with the customers paying for the product, which is a position with leverage when negotiating internally or considering founding. The downside is that FDE comp at the senior IC tier is harder to verify publicly than equivalent SWE comp, because (a) the role is newer, (b) levels.fyi has fewer FDE samples than SWE samples, and (c) much of the public material is content marketing. Don't assume FDE pays more or less than SWE at the same company without verified data.

Career paths beyond 'go found a company'

The founder pipeline is documented but small. What our research couldn't pin down with high confidence is the relative size of other paths: ex-FDEs who become PMs, who become EMs leading multiple FDE pods, who become Partners at customer-facing VC firms (the role has obvious overlap with the work of an enterprise investor), who become founding engineers at other startups rather than founding themselves, or who go to other AI labs. All of these paths exist in LinkedIn searches; none is rate-quantified in a source we trust. The takeaway is that the role is structurally exit-friendly (lots of paths) but specifically 'FDE -> founder' is the only path with named, funded examples we could verify. If your decision hinges on knowing 'what fraction of ex-FDEs found companies,' the honest answer is that no source we verified gives a credible base rate.

What to do with all this if you're deciding

If you're using comp to choose between roles, pick the role whose base is high enough that you don't feel pressured into the daily work, then weight the equity comp by how much you trust the company's outcome. Anthropic's $200K-$300K base is solid floor regardless of equity; Palantir's median $215K is the verified reference point for everyone else. If you're using career trajectory to choose, the verified picture is: founding is a real but small path; PM/EM/Partner/founding-engineer paths exist but rate isn't documented; staying IC at frontier AI labs is plausibly attractive because the work is genuinely interesting and the demand isn't going away. The role is not a guaranteed founder accelerator. It is a high-leverage seat that some people convert into founding and others convert into something else.

Key takeaways

•Palantir FDSE comp (Levels.fyi, June 2026): median $215K, range $171K-$415K+. Use this as the reference point.
•Anthropic's posted FDE base is $200K-$300K. They have a no-negotiation policy. Quote: 'We don't negotiate that level because we think it's unfair' (Dario Amodei, Fortune Aug 2025).
•Non-Palantir total-comp figures circulating online (OpenAI $1.0M-$1.28M, Hashnode bands, Secondtalent bands) failed adversarial verification. We don't quote them here, and you shouldn't either.
•Founder pipeline is real: Miliauskas → Mavenoid ($40M), McNeal → Thread AI ($26M), Ural/Ranade/Speare → Arena ($32M). 40+ ex-FDE-founded companies is the conservative subset cited.
•We do not have verified base rates for what fraction of ex-FDEs end up as founders vs. PMs vs. EMs vs. staying IC. Don't trust round numbers about this.
•If choosing on comp, weight the verified base over speculative total-comp claims. If choosing on trajectory, treat the role as exit-friendly, not as a guaranteed founder track.

Exercise

Open Levels.fyi and look up the FDE / Applied AI Engineer titles at the specific company you're targeting. Note the sample size (how many data points). If it's under ~20, treat the median as anecdote, not signal. If your target company is OpenAI or Anthropic, also pull base comp from their own published listing on the careers page. Triangulate.

Self-check

1.Quote the Palantir FDSE comp range from Levels.fyi and explain why that number is more trustworthy than the OpenAI L5/L6 figures floating around online.
2.Name three verified ex-Palantir FDE-founded companies with funding totals.
3.Why does Anthropic's no-negotiation policy structurally change how you'd evaluate the role's base?
4.If a friend says 'half of all ex-FDEs become founders,' what's the honest pushback?

Sources

Next lesson

Failure Modes and the Honest Case Against

The org-level trap that turns FDE teams into consulting shops, the misappropriation problem the role has on LinkedIn, and the personal-burnout dimension we deliberately won't fake.

→