2026 Software Engineer Salary & Hiring Report: 1,407 Jobs Analyzed

An original study by roasted.cv. Data collected June 2026. Free to cite with attribution and a link to roasted.cv.

We pulled 1,407 live, de-duplicated software engineer job postings from Google's jobs index across ten major US metros plus national searches, then read the full text of 201 of them to measure what employers actually ask for. No surveys, no self-reported salaries, no recycled Bureau of Labor Statistics tables. Just the listings that were open in the second week of June 2026.

Here is what the market looked like.

Key findings

The median disclosed pay range midpoint was $170,000. Half of all postings that listed pay fell in a $140K to $200K band. The top of the market (OpenAI, HubSpot, Lambda, Perplexity, xAI, Snowflake) posted ranges topping out at $465K to $500K.
Python and machine learning are now tied as the most-requested capabilities, each named in 50% of job descriptions. Java (42%) and "system design / distributed systems / scalability" (41%) follow. AWS (39%) is the single most-named tool after languages.
Remote is rarer than the discourse suggests. When you search a specific city, only 6.5% of local listings were explicitly remote. Across all searches including a dedicated remote query, the figure rises to 18%, but location-based job hunting still returns an overwhelmingly onsite market.
This is a senior's market. 57% of postings targeted senior, staff, principal, lead, or manager levels. Only 8.5% were junior, entry, or internship roles.
Geography still sets a $79K spread. Median midpoints ran from $215K in San Francisco and $214K in Seattle down to $136K in Atlanta.
Frontend pays the least of the major specializations ($154K median); data, platform, and ML roles pay the most ($176K to $198K).
92.9% of postings were full-time. Contract, part-time, and internship roles combined for under 8%.
The biggest single source of postings was LinkedIn (19% of all listings), followed by Indeed (7.8%). Defense and aerospace employers (Lockheed Martin, Boeing, RTX, Northrop Grumman, Anduril) appeared as often as big tech.

1. Salary: the disclosed market clusters tightly around $170K

Of the 1,407 postings, 371 (26%) disclosed a pay range. We took the midpoint of each range and built the distribution below. (US salary-transparency laws mean disclosure is concentrated in California, Washington, New York, Colorado, and Illinois and among larger employers, so these figures describe the disclosed market, which sits above the true national median. See methodology.)

Table 1: Disclosed salary distribution (annualized range midpoints, n = 371)

Percentile	Annual pay (midpoint of posted range)
Minimum	$36K
10th	$112K
25th	$140K
Median (50th)	$170K
75th	$200K
90th	$240K
Maximum	$419K

The typical posted range itself ran from a median floor of $140K to a median ceiling of $200K. The market is tight in the middle and long-tailed at the top: the highest-paying postings were almost all frontier AI labs and high-growth infrastructure companies.

Table 2: Median pay by metro (range midpoints; general "software engineer" search per city)

Metro	Postings	Disclosed	Median midpoint	Explicitly remote
San Francisco, CA	98	26	$215K	9%
Seattle, WA	99	18	$214K	7%
New York, NY	95	24	$179K	9%
Denver, CO	99	23	$176K	13%
Washington, DC	98	15	$173K	6%
Los Angeles, CA	100	28	$170K	1%
Chicago, IL	100	30	$155K	2%
Boston, MA	100	18	$151K	3%
Austin, TX	98	12	$150K	7%
Atlanta, GA	96	17	$136K	7%

San Francisco and Seattle sit roughly $60K to $80K above Atlanta, Austin, and Boston. The Bay Area premium survived every layoff cycle since 2022.

Table 3: Median pay by seniority (n = 371 disclosed)

Level	Postings disclosed	25th	Median	75th
Staff / Principal	59	$190K	$217K	$253K
Lead / Manager	27	$160K	$185K	$210K
Senior	126	$144K	$165K	$200K
Mid / Unspecified	121	$138K	$165K	$191K
Mid (level II)	13	$123K	$140K	$169K
Junior / Entry	24	$76K	$100K	$114K

The jump from entry to staff more than doubles pay: a $100K median for juniors against $217K for staff and principal engineers.

Table 4: Median pay by specialization (from job titles, n = 371 disclosed)

Specialization	Postings disclosed	Median midpoint
Data engineering	10	$198K
Platform / DevOps / Infra	28	$184K
Mobile	7	$179K
Machine learning / AI	61	$176K
Backend	39	$175K
Full-stack	16	$165K
General software engineer	157	$160K
Frontend	37	$154K
Embedded	13	$135K

Frontend and embedded roles anchor the bottom; data, platform, and ML roles command the premium. (Mobile and data cells have small samples; treat as directional.)

2. Skills: Python and machine learning now lead, with AWS as the default cloud

We fetched and parsed the full job-description text of 201 postings and counted how often each technology was named. A skill counts once per posting regardless of repetition.

Rank	Skill	% of postings
1	Python	50%
2	Machine learning / deep learning	50%
3	Java	42%
4	System design / distributed systems / scalability	41%
5	AWS	39%
6	React	31%
7	Kubernetes	29%
8	Agile	29%
9	JavaScript	28%
10	CI/CD	27%
11	C++	26%
12	TypeScript	23%
13	Git	23%
14	REST APIs	22%
15	SQL	21%
16	Docker	21%
17	Microservices	20%
18	GCP	19%
19	Linux	18%
20	Azure	18%
21	LLMs / generative AI	18%
22	.NET	16%
23	Node.js	13%
24	Angular	12%
25	Spring	12%
26	Kafka	12%
27	Go	11%
28	C#	9%
29	Terraform	9%
30	PostgreSQL	8%

Three things stand out:

Machine learning has gone mainstream. It is named as often as Python and twice as often as it would have been on a generalist software engineer listing five years ago. A named requirement for LLMs or generative AI appeared in 18% of postings, and a reference to AI or ML somewhere in the text (including company boilerplate) appeared in 97%.
Cloud is a baseline, and it is AWS. At least one of AWS, Azure, or GCP was named in 43% of descriptions, with AWS (39%) far ahead of GCP (19%) and Azure (18%).
"System design" is the most-requested non-language skill. Architecture and distributed-systems thinking (41%) outranked every individual framework. The interview obsession is real in the job text too.

3. Remote work: scarcer than the headlines

The remote-vs-office debate gets framed as if half of all engineering jobs are remote. The listings disagree.

Table 6: Work arrangement

Arrangement	Measurement
Explicitly remote, metro-local searches	6.5% (64 of 983 city-targeted listings)
Explicitly remote, all searches incl. dedicated remote query	18% (258 of 1,407)
Hybrid (explicitly labeled)	1.6%
Onsite or unspecified location	~80%

When a real job seeker searches "software engineer" in their city, only about 1 in 15 results is openly remote. The Denver-to-LA range was 1% to 13%. Remote roles exist, but they are concentrated, competitive, and do not dominate the local market the way remote-first commentary implies.

4. Who the market wants: senior, full-time, and increasingly specialized

Table 7: Seniority mix (from job titles, n = 1,407)

Level	Postings	Share
Senior	451	32.1%
Mid / unspecified	439	31.2%
Staff / Principal	228	16.2%
Lead / Manager	124	8.8%
Junior / Entry	107	7.6%
Mid (level II)	46	3.3%
Intern	12	0.9%

Senior, staff, lead, and manager titles made up 57.1% of all postings. Junior, entry, and internship roles combined for 8.5%. The "no entry-level jobs" complaint that floods engineering forums has a basis in the listing data.

Table 8: Contract type (n = 1,407)

Type	Postings	Share
Full-time	1,307	92.9%
Contractor	47	3.3%
Full-time and part-time	24	1.7%
Internship	16	1.1%
Part-time	4	0.3%
Full-time and contractor	4	0.3%

Table 9: Specialization signal (from job titles; titles can carry more than one tag, n = 1,407)

Specialization	Postings	Share of all
Machine learning / AI	212	15.1%
Platform / DevOps / Infra	140	10.0%
Backend	118	8.4%
Frontend	110	7.8%
Full-stack	77	5.5%
Data	50	3.6%
Embedded	42	3.0%
Mobile	20	1.4%
QA / Test	18	1.3%
Security	13	0.9%

ML/AI is now the single most common specialization called out in titles, ahead of backend and frontend. The generalist "Software Engineer" title is still the most common label overall, but where employers do specialize, they increasingly specialize toward AI and platform work.

5. Who is hiring, and where the jobs live

Table 10: Most active employers (by posting count in sample)

Employer	Postings	Sector
Lockheed Martin	25	Defense / aerospace
NVIDIA	16	AI hardware
Capital One	16	Finance
Boeing	16	Aerospace
Waymo	15	Autonomous vehicles
RTX (Raytheon)	14	Defense
Northrop Grumman	14	Defense
General Motors	13	Automotive
Cox Automotive	11	Automotive
Snowflake	11	Data infrastructure
Google	9	Big tech
Anduril Industries	9	Defense tech
Pinterest	9	Big tech
Adobe	9	Big tech
Apple	8	Big tech

The surprise for many readers: defense and aerospace employers were as visible as Silicon Valley. Five of the fifteen most active hirers build weapons systems, aircraft, or defense software. Autonomous-vehicle and automotive software (Waymo, GM, Cox) formed a second cluster, and AI infrastructure (NVIDIA, Snowflake) a third. (Note: a handful of high-volume accounts in the raw data were staffing agencies and bootcamp recruiters rather than direct employers; those are excluded from this table.)

Source	Share
LinkedIn	19.0%
Indeed	7.8%
ZipRecruiter	4.0%
Workday (company ATS)	3.3%
Welcome to the Jungle	2.8%
BeBee	2.3%
Dice	1.1%
Teal	1.0%
Company career sites (combined)	the remainder

LinkedIn is the center of gravity, carrying nearly one in five postings. But the long tail matters: a large share of listings appear only on company applicant-tracking systems (Workday, Greenhouse-style pages, and direct career portals), which is exactly the inventory that generic job-board aggregators miss.

Freshness

Of the 910 postings that carried a visible date, 46% were posted within the previous 7 days and 30 were posted within 24 hours. This is a live, fast-cycling market: the median listing in our sample was days, not weeks, old.

What this means if you are job hunting in 2026

Lead your resume with Python, a cloud (AWS first), and one concrete machine-learning or distributed-systems project. Those four signals match what half the market is screening for. A resume that buries them under a wall of frameworks is optimizing for the wrong filter.
Name "system design" experience explicitly. It is the most-requested non-language skill and the easiest to leave implicit and lose credit for.
If you are early-career, widen your net beyond your city. Entry-level roles are 8.5% of the market and remote roles are 6.5% of local results, so the two scarcest segments rarely overlap. Apply nationally and to defense, automotive, and enterprise employers, not just consumer tech.
If you are senior, the leverage is in specialization. Data, platform, and ML titles paid $176K to $198K at the median against $154K for frontend. The premium for moving toward infrastructure and AI is real and measurable.
Apply through the company ATS when you can. A meaningful slice of inventory never reaches the big aggregators. The career-site listings are less crowded.

Methodology

Source. Job postings were collected from Google's jobs aggregator. This indexes listings syndicated to Google from LinkedIn, Indeed, ZipRecruiter, company ATS platforms, and hundreds of niche boards.

Queries. Sixteen searches run on June 15, 2026, in English, US market: a general "software engineer" search in each of ten metros (New York, San Francisco, Seattle, Austin, Chicago, Boston, Los Angeles, Atlanta, Denver, Washington DC), plus six national searches ("senior software engineer", "entry level software engineer", "software engineer remote", "backend engineer", "frontend engineer", "machine learning engineer") to ensure coverage of seniority bands, specializations, and remote inventory.

De-duplication. Raw results (1,527 items) were de-duplicated on Google's job id, yielding 1,407 unique postings.

Salary. 371 postings (26%) disclosed a pay range. Hourly and monthly figures were annualized (hourly x 2,080; monthly x 12); comma- and "K"-formatted numbers were normalized; the midpoint of each range was used for distribution statistics. Disclosure skews toward salary-transparency states (CA, WA, NY, CO, IL) and larger employers, so reported medians describe the disclosed market and run above the true national median for the role.

Skills. Full job-description text was fetched directly from source URLs for 201 postings (the share of source pages that served readable HTML). Each posting was matched against a dictionary of 46 technologies using word-boundary patterns. A skill is counted once per posting. The parsed corpus skews somewhat toward defense, enterprise, and startup listings; treat the skill percentages as directional rather than exact.

Seniority and specialization were classified from job titles via keyword rules. Work arrangement was inferred from the location field ("Anywhere", "Remote") and title text.

Limitations. This is a point-in-time snapshot of one country and one role family from one aggregator. It measures demand as advertised, not hires. Salary disclosure is partial and geographically skewed. Skill frequencies depend on which source pages were machine-readable. Figures are presented to support directional conclusions, not payroll-grade precision.