All reports
software engineering peer pack · five company records · 2026

Five Software Engineer (AI-native companies) benches on one date: inside the 818-profile peer pack

Scale AI leads at 364 (44.5% of the 818-person pack in the AI-native companies cut) against Glean at 228. As of 2026-10-29 PT.

As-of 2026-10-29 PT Published by Metix AI Collection series
Who holds how many on this seat?

Who holds how many on this seat?

The ledger opens with Scale AI at 364, as of 2026-10-29 PT.

CompanyCurrentShare of pack
Scale AI36444.5%
Glean22827.9%
OpenAI11814.4%
Harvey728.8%
Hugging Face364.4%
Named list

Get the named list behind these 818 Software Engineers

Take the named list for the employers shown here. Keep the request inside the companies and titles already shown.

Which host is the pack anchor?

Which host is the pack anchor?

The largest bench is Scale AI at 364/818 (44.5%).

Scale AI Glean OpenAI Harvey Hugging Face 0 100 200 300 400 364 228 118 72 36 Software Engineer peer pack profiles · As of 2026-10-29 PT Scale AI Glean OpenAI Harvey Hugging Face 0 100 200 300 400 364 228 118 72 36 Software Engineer peer pack profiles · As of 2026-10-29 PT Software Engineer bench contrast, highest to lowest: Scale AI 364, Glean 228, OpenAI 118, Harvey 72, Hugging Face 36. As of 2026-10-29 PT. Source: Metix AI.
Software Engineer bench contrast, highest to lowest. As of 2026-10-29 PT. Source: Metix AI.
Data & notes
Scale AI 364, Glean 228, OpenAI 118, Harvey 72, Hugging Face 36.
NameValue
Scale AI364
Glean228
OpenAI118
Harvey72
Hugging Face36
Named list

Compare the five benches

The named list covers the companies and titles on this page. Keep the title family this page counted.

Key takeaway · How to cite
0% 20% 40% 60% 80% 100% 44.5% 27.9% 14.4% Scale AI 364 (44.5%) Glean 228 (27.9%) OpenAI 118 (14.4%) Harvey 72 (8.8%) Hugging Face 36 (4.4%) Software Engineer pack share profiles · As of 2026-10-29 PT 0% 30% 60% 90% 44.5% 27.9% 14.4% Scale AI 364 (44.5%) Glean 228 (27.9%) OpenAI 118 (14.4%) Harvey 72 (8.8%) Hugging Face 36 (4.4%) Software Engineer pack share profiles · As of 2026-10-29 PT Software Engineer pack share of 818: Scale AI 364 (44.5%), Glean 228 (27.9%), OpenAI 118 (14.4%), Harvey 72 (8.8%), Hugging Face 36 (4.4%). As of 2026-10-29 PT. Source: Metix AI.
Software Engineer pack share of 818. As of 2026-10-29 PT. Source: Metix AI.
Data & notes
Scale AI 364 (44.5%), Glean 228 (27.9%), OpenAI 118 (14.4%), Harvey 72 (8.8%), Hugging Face 36 (4.4%).
NameValueShare
Scale AI36444.5%
Glean22827.9%
OpenAI11814.4%
Harvey728.8%
Hugging Face364.4%
Headcount vs share: who leads?

Absolute headcount vs share — who leads? Anchor Scale AI

Absolute headcount and share % can diverge: who is largest versus who holds the pack.

0 100 200 300 400 Scale AI Glean OpenAI Harvey Hugging Face 364 228 118 72 36 44.5 27.9 14.4 8.8 4.4 Headcount Pack share % Software Engineer: headcount vs pack share % As of 2026-10-29 PT 0 100 200 300 400 Scale AI Glean OpenAI Harvey Hugging Face 364 228 118 72 36 44.5 27.9 14.4 8.8 4.4 Headcount Pack share % Software Engineer: headcount vs pack share % As of 2026-10-29 PT Absolute headcount beside pack share % for Software Engineer. Anchor Scale AI 364 (44.5%); tail Hugging Face 36 (4.4%). As of 2026-10-29 PT. Source: Metix AI.
Absolute headcount beside pack share % for Software Engineer. As of 2026-10-29 PT. Source: Metix AI.
Data & notes
Anchor Scale AI 364 (44.5%); tail Hugging Face 36 (4.4%).
CategoryHeadcountPack share %
Scale AI36444.5
Glean22827.9
OpenAI11814.4
Harvey728.8
Hugging Face364.4
How concentrated is the pack?

How concentrated: top two / mid / tail?

Top two hold 592/818 (72.4%); mid 190; tail 36.

0% 20% 40% 60% 80% 100% 72.4% 23.2% Scale AI+Glean 592 (72.4%) OpenAI+Harvey 190 (23.2%) Hugging Face 36 (4.4%) Software Engineer concentration (top2 / mid / tail) profiles · As of 2026-10-29 PT 0% 30% 60% 90% 72.4% 23.2% Scale AI+Glean 592 (72.4%) OpenAI+Harvey 190 (23.2%) Hugging Face 36 (4.4%) Software Engineer concentration (top2 / mid / tail) profiles · As of 2026-10-29 PT Concentration of the 818-person Software Engineer pack: top two Scale AI+Glean = 592 (72.4%); mid OpenAI+Harvey = 190 (23.2%); tail Hugging Face = 36 (4.4%). As of 2026-10-29 PT. Source: Metix AI.
Concentration of the 818-person Software Engineer pack. As of 2026-10-29 PT. Source: Metix AI.
Data & notes
top two Scale AI+Glean = 592 (72.4%); mid OpenAI+Harvey = 190 (23.2%); tail Hugging Face = 36 (4.4%).
NameValueShare
Scale AI+Glean59272.4%
OpenAI+Harvey19023.2%
Hugging Face364.4%
How to use it

What it means for Software Engineer (818) / how to use it

This week open the named-list ask on Scale AI: 364/818 (44.5%) — the largest bench first.

What is in-scope for this seat?

What is in-scope for this seat?

Records

Records

Each number comes from one company's own exclusive homepage record: Scale AI (www.scale.com), Glean (www.glean.com), OpenAI (www.openai.com), Harvey (www.harvey.ai), Hugging Face (www.huggingface.co).

As of 2026-10-29 PT · Source: Metix AI

Notes & method

Methodology

The count is current Software Engineer profiles on one company's exclusive homepage record at 2026-10-29 PT. The pack total adds five non-overlapping company records. Metix AI is the source.

Aggregate report · no personal information shown · provided by Metix AI
FAQ

FAQ

Why these five companies for the Scale AI-led Software Engineer pack (818)?
Software Engineer is counted on five separate company records. The published total is 818, dated 2026-10-29 PT.
Are the other titles in the Software Engineer total of 818?
The total excludes Engineering Manager and Site Reliability Engineer. 2026-10-29 PT and 818 refer to Software Engineer alone.
Why do the two ends of the Software Engineer pack (818) differ?
Both ends use current Software Engineer on each company's own record. The date is 2026-10-29 PT, so the gap is bench depth.
Can this Software Engineer seat (818) be reused?
The published 818 stays fixed at 2026-10-29 PT. The Software Engineer wording is reusable beyond these five.

Request the software engineering peer pack (five names · 818 current people)

Ask for people on the measured wires, not a broader uncut list. The request should name the same employers.

Optional: try Mira
中文版