Oracle's 436 LCAs vs 21,000 Cuts: A Resume Play for the Backfill

The Blind thread going around this week ("Slap in our face: Oracle reportedly laying off 10k+ employees, while still filing H-1B petitions and LCAs") is loud for a reason. Oracle's US headcount fell from roughly 162,000 to 141,000 in fiscal 2026, and the company kept filing hundreds of Labor Condition Applications through the cuts. That contradiction is uncomfortable, but for anyone who was pushed out, it is also the most useful public dataset you have.

What the Oracle layoffs + H-1B story actually is

Oracle cut roughly 21,000 net US roles in fiscal 2026 while filing 436 H-1B petitions in the same year, mostly to fund an AI infrastructure buildout, not to shrink the company. The Form 10-K logs $1.84 billion in restructuring charges, almost all severance, alongside $55.7 billion spent on AI infrastructure and $43 billion in new borrowing to help cover it. Oracle Cloud Infrastructure revenue grew 77% to $18.1 billion in the same year.

Read that sequence carefully: this is not "10,000 out, 10,000 in." It is a company trading generalist headcount for GPU capacity and a narrow set of specialists. The 436 H-1B filings against 21,000 net departures work out to a backfill rate of about 15%, and even that overstates it because LCAs and petitions can be renewals for people who already work there. The Department of Labor explicitly notes that its disclosure data does not indicate the employer's intended use for each LCA.

That precision matters. The Oracle layoffs H-1B story is not a scandal of one-for-one replacement. It is a shortlist of exactly which specialties Oracle still values enough to sponsor, published in a searchable format by the federal government.

Why public LCA filings are the highest-signal job board you're not using

Public LCA filings are the certified job offers US employers must file with the Department of Labor before they can hire, extend, or transfer an H-1B worker, and each row publishes the exact title, worksite, and wage the employer committed to on paper. That is more specific than any job posting you will see on LinkedIn.

Here is what a single row on an LCA disclosure contains and how to weaponize each field:

• Employer name and FEIN. Confirms the legal entity, not the marketing brand. Oracle America, Inc. and Oracle Cerner are different filers.
• SOC code. Tells you the government's official job family. Match this to your own resume's target family, not the flowery title.
• Job title. The company's internal ladder title ("Principal Member of Technical Staff," "Applications Developer 4"). This is the phrase to mirror in your headline.
• Wage level (I through IV). Level I is entry, Level IV is expert. This is the single highest-signal field on the form.
• Prevailing wage source and offered wage. The floor Oracle told the government it would pay.
• Full-time or part-time status.
• Worksite geography. City, county, state. Match this to your own commute or remote radius.

Four places index these filings for free:

1. h1bdata.info (over 4.8 million records indexed between Oct 2013 and Sept 2025)
2. h1bgrader.com/lca-search
3. h1bdatawatch.com
4. The OFLC's own quarterly disclosure CSVs (raw, ugly, complete)

One caveat before you rebuild your resume around a specific wage: certified filings represent job offers, not actual hires or final compensation. The base wage excludes stock, bonuses, and sign-on. Treat the number as a published floor, not a paycheck.

The Wage Level I to IV trick almost no job seeker uses

Wage Level is the field to build your resume around, because it tells you where on the seniority ladder Oracle currently has open sponsorship budget, and it lets you translate that floor into a negotiation number at every competitor within commuting distance.

Think about the mechanics. If Oracle files 40 LCAs for "Principal Member of Technical Staff" at Level IV in Seattle at a specific wage, three things are now true:

1. That title exists in real volume at Oracle Seattle.
2. Oracle has committed on the record to paying at least that base for that title.
3. Any Amazon, Microsoft, or Snowflake recruiter two miles away is competing against that same floor.

Sort your LCA search by Level IV and your own metro. The titles that repeat are the ones where the company has a genuine specialist shortage, not a re-org shuffle. Those are the titles a laid-off senior should mirror in the top two inches of the resume. The wage figure is your opener when a recruiter at a competitor asks about compensation expectations.

This is also the exact place Refolk earns its keep. Paste the LCA title and the specific job posting you found at the competitor, and Refolk rewrites your resume against both: the title language the ATS is scanning for, and the seniority signals implied by the wage level. You get back your own history reframed to intercept the search that recruiter is actually running.

The OCI trap: why Oracle stacks do not translate on their own

Oracle-specific stacks (OCI, Exadata, Fusion, NetSuite, Cerner) need to be translated into cloud-neutral, AWS/Azure/GCP-friendly impact statements or they get filtered out before a human sees them, because recruiters do not search for "OCI." In Refolk's index of professional profiles, only 19 US engineers list OCI as a named skill. Seven of them still work at Oracle. Two are at Google, and one each sit at Apple, Red Hat, Datadog, and Microsoft. That is the entire visible OCI talent graph outside Oracle: roughly a dozen people.

The AWS side of the same graph tells the other half of the story. Refolk's index shows 29 SWE, Sr SWE, and Cloud Engineer profiles that list AWS as a skill and have "Oracle" in the headline. Seventeen still work at Oracle. The rest have landed at Snap, Goldman Sachs, SoFi, Microsoft AI, and Fidelity. That is the ex-Oracle absorption cohort, and every one of those engineers rewrote OCI experience into AWS-legible bullets to land the next role.

The 21-of-25 number is the punchline. If you keep "OCI" as your marquee skill, you are competing for a tiny number of Oracle-badged seats. If you translate it, the addressable market opens up.

Translation, concretely:

• "Migrated 40TB workload from on-prem Exadata to OCI Autonomous Database" becomes "Migrated 40TB OLTP workload to a managed cloud database (Oracle Autonomous DB on OCI, equivalent pattern to AWS Aurora / Azure SQL Managed Instance), cutting query latency 38%."
• "OCI Object Storage lifecycle policies" becomes "S3-equivalent object storage lifecycle policies (implemented on OCI)."
• "Fusion Middleware integration" becomes "REST/SOAP service integration across ERP (Oracle Fusion), roughly analogous to MuleSoft or AWS AppFlow deployments."

Every one of those is a bullet a recruiter can search for. This tailoring is exactly what Refolk does per posting: the OCI-heavy version stays for Oracle recompete roles, and a cloud-neutral version gets generated the moment you paste an AWS or GCP job description.

How to actually mine LCA filings in an afternoon

Here is the workflow that turns public LCA data into a targeted application list in a single sitting.

1. Search your former employer first. Pull every certified LCA Oracle filed in the last four quarters at your worksite. Note repeating titles and their wage levels. These are the job families Oracle still wants filled, i.e. the roles your internal reputation and reference network still translate into cleanly.
2. Cross-shop the same SOC code at three competitors. Amazon filed roughly 2,675 H-1B petitions in the same window; Microsoft publishes thousands of LCA records annually and is the standard tech benchmark. Pull the same SOC code and wage level at both. You now have three comparable wage floors for the same work.
3. Filter by worksite geography, not company. LCA data is worksite-specific. If you can only work in Austin or Bellevue, sort every filer by county. This surfaces mid-cap employers that pay Level IV wages but never trend on Blind.
4. Match titles to your own history, then rewrite. The ATS scans for the exact title strings the LCA uses. If Oracle's LCA says "Applications Developer 4" and Amazon's says "Software Development Engineer II," and both are the same SOC + wage level, you want the competitor's title in your headline when you apply there.
5. Do the WARN math in parallel. A WARN Act filing confirms Oracle separations were expected by June 1, 2026. If Oracle skipped the required 60-day heads-up for a qualifying site, you may be owed 60 days of back pay on top of severance, though Oracle may argue remote-worker classification exempts certain sites. Pair this with an LCA search on the same worksite address to show the site was staffed on paper the day you were cut.

Step 4 is where most people stall, because rewriting the same resume against ten different LCA titles is dull, mechanical work. It is also the exact work Refolk takes off you: paste the posting (or the LCA row), get your own resume back rewritten for it, plus a cover letter draft and a fit score that tells you whether the application is worth sending in the first place.

The urgency layer: grace period, WARN, and the IBM pattern

If you are on an H-1B and were laid off, the 60-day grace period may not exist by the time you finish reading this article, which is why LCA data is the shortest path to a shortlist of sponsors who have already proven they will file. DHS/USCIS submitted a proposed rule titled "Eliminating the Discretionary 60-day Grace Period" to OMB on August 6, 2026. If it lands, an H-1B holder who is separated no longer has two months to find a new position.

Mining LCA disclosure data compresses the sponsor-vetting step from weeks to hours. You are no longer asking "does this company sponsor?" You are reading their filings.

The Oracle situation is also not unique. IBM's 2026 layoffs cut roughly 9,000 US positions while tripling entry-level hiring and shifting experienced roles to India. Nina Lewis, a 34-year Oracle veteran, wrote in a LinkedIn post that the Oracle cuts appeared to follow "an algorithm" targeting high-level contributors and mid-level managers with outstanding stock options. Whether or not that read is right, the observed pattern (senior, expensive, vested) matches what the peer companies are doing.

The takeaway is not "Oracle betrayed you." The takeaway is that senior US tech workers can now read exactly which titles, wage levels, and worksites the largest sponsors still care about, and rewrite one resume per target to intercept that specific opening. The data is public. The 60-day clock might not be.

FAQ

Where do I find public LCA filings for a specific company?

Search h1bdata.info, h1bgrader.com/lca-search, or h1bdatawatch.com by employer name. All three index the Department of Labor's certified LCA disclosures for free, and h1bdata.info alone holds more than 4.8 million records from Oct 2013 through Sept 2025. For raw completeness, download the OFLC's own quarterly disclosure CSVs. Filter by worksite state and certification date to isolate current openings from historical filings.

Does an LCA filing mean the company is actually hiring someone new?

Not necessarily, which is the single most important caveat in this whole exercise. LCAs and H-1B petitions can be renewals or extensions for employees who already work there, and DOL data does not indicate the employer's intended use. Treat LCA volume as a signal that a job family is active and sponsored, not proof of a specific open req. Cross-check against the company's public careers page or a recruiter conversation before you assume a seat is open.

How do I use the Wage Level field on a resume?

Use it to reverse-engineer the seniority band the employer expects for a title, then mirror that title and its language in your resume headline and top bullets. Level I is entry, Level IV is expert. If competitors in your metro file the same SOC code at Level IV with a specific wage floor, you now have both a title target and a compensation floor to negotiate against. The wage figure excludes stock and bonus, so treat it as a base minimum.

What does Refolk do with an LCA row?

Paste the LCA title, wage level, and worksite (or the matching job posting) into Refolk, and it rewrites your resume against that specific target: mirroring the title language recruiters and ATS actually search for, translating Oracle-specific stack terms (OCI, Fusion, Exadata) into cloud-neutral equivalents where the posting expects them, and drafting the cover letter. Refolk also scores how well your history actually fits the role, so you skip the applications that were never going to convert.