Citi's Sept 10, 12, 30 NYC WARN: Three Backend Drops Nobody Is Sourcing

Citigroup posted three NYC WARN closure dates to the New York DOL dashboard on August 6, 2026: September 10, September 12, and September 30. Every AI-focused recruiter in Manhattan is going to filter these people out because their resumes say "bank." That is the arbitrage.

What Citi actually filed, and why the dates matter

Citi filed three separate NYC WARN closure dates - 9/10, 9/12, and 9/30 - tied to the latest tranche of Jane Fraser's plan to eliminate 20,000 roles globally. Because New York's mini-WARN requires 90 days advance notice (versus 60 federal) and applies to any employer with 50+ employees, everyone on those lists has known since early June 2026 and has been on payout for a full quarter.

Three things follow from that timing:

• The candidates are not in shock on the closure date. They have spent 90 days on payroll, on notice, and quietly interviewing.
• The three dates are not one event. They almost certainly map to different Citi legal entities or business units, which means three different cohorts, not one duplicate list.
• The 90-day payout window ends on the WARN date, so financial pressure to accept an offer peaks on 9/11, 9/13, and 10/1 respectively.

Citi's HQ at 388 Greenwich Street in Tribeca is the physical anchor for all three filings, per Crain's. The internal code name for the restructuring, first reported by CNBC, is Project Bora Bora. Both details are useful in outreach because they signal you actually know what happened to them.

Why the "bank engineer" filter is a $0 arbitrage

The reason most AI-focused sourcers will skip this pool is a stereotype: bank equals COBOL, mainframes, and change-review boards. That stereotype is exactly wrong for the 2026 Citi cut, and the mismatch is what makes this window valuable.

Citi's Transformation program, the internal name for the modernization effort, is projected to yield roughly $2.5 billion in cost savings and is explicitly aimed at upgrading aging technology. The engineers being cut are not the mainframe holdouts. They are the people who spent the last two years replacing the mainframe. Fraser said in January 2026 that more than 80% of the Transformation effort is complete, which means the workstream is winding down and the workstream engineers are being released.

Fraser's own memo attributes the cuts to automation and AI: "some roles will change, new ones will emerge and others will no longer be required." Translation: the engineers on the 9/10, 9/12, and 9/30 lists built the systems that are now replacing them. They know it. They are not defensive about it.

69%

Where Citi engineers actually sit

In Refolk's index of professional profiles, a US-wide sample of 25 currently-in-role Citi engineers and developers concentrates in three metros: NYC, Texas (Dallas, Frisco, Fort Worth, and Flower Mound), and Tampa. NYC is the only one of the three with a September WARN filing on the board.

Two implications from that distribution:

1. If you are sourcing outside NYC, WARN gives you no signal. You have to work the LinkedIn and GitHub trail directly.
2. The NYC pool is not gigantic. It is concentrated, which is exactly what makes it sourceable in a week rather than a quarter.

This is the exact gap Refolk closes for the non-NYC pieces: you describe the engineer you want ("ex-Citi platform engineer, Kafka or Kubernetes background, currently in Tampa or Dallas") in plain English and get a ranked shortlist across GitHub, LinkedIn, and the open web, without having to know in advance which subsidiary they sat in.

The "VP" trap that kills half of these searches

Citi "Vice President" is a rank-and-file engineering title, not an executive one. If you search LinkedIn for "Software Engineer" at Citigroup and stop there, you miss the entire senior individual contributor layer, and Refolk's index confirms VP appears throughout the engineering pool.

The Citi engineering title ladder that a sourcer needs to search on:

• Analyst (junior)
• Associate (mid)
• Assistant Vice President, or AVP (senior)
• Vice President, or VP (staff-equivalent IC or engineering manager)
• Senior Vice President, or SVP (principal-equivalent or director)
• Director / Managing Director (executive)

Fraser's restructuring collapses management from 13 layers to 8. That compression falls hardest on the middle three rungs: AVP, VP, and SVP. Those are precisely the profiles a Series A or B founder wants for a first platform hire. If your Boolean string does not include "Vice President" as a target, you are self-selecting out of the best half of the list.

Citi VPs are not executives. They are the staff engineers you have been trying to poach from Stripe for eighteen months.

Three dates, three campaigns, not one blast

The single biggest tactical mistake sourcers will make on this pool is treating 9/10, 9/12, and 9/30 as one event. They are not. The three dates almost certainly correspond to three different Citi legal entities or business units filing separately, which means three different cohorts of ex-Citi platform engineers with different skill mixes.

Sequence your outreach:

1. Week of 9/8: Warm the 9/10 cohort. Reference their end date directly. Offer a first conversation before their payout ends.
2. Week of 9/10: Hit the 9/12 cohort with a different opener. Do not recycle the 9/10 message; a fraction of the 9/12 list will have already seen it forwarded internally.
3. Week of 9/28: Hit the 9/30 cohort. This is typically the largest tranche and the one everyone else will have burned their message on.

Between the tranches, verify each candidate individually. A candidate on the 9/12 list may already have accepted an offer sourced from the 9/10 blast. This is where a plain-English query beats a static list: instead of maintaining a spreadsheet, you can ask for "ex-Citi VPs in NYC who left around 9/12 and have shipped Kafka in the last 24 months" and get a live answer.

The AI-elimination reframe that changes your pitch

Fraser publicly attributed the cuts to AI and automation, and every candidate on these lists read the memo. That public framing is your best pitch angle, because the candidates already know their old job was killed by AI and they want to be on the offensive side of the trade next.

The wrong pitch: "Why did you leave a bank?" Defensive, and it wastes their 30 minutes.

The right pitch, in this order:

• Acknowledge the Transformation program by name. It signals you did the reading.
• Reference the specific WARN date they are on. It signals you understand the timeline pressure.
• Frame the role as building the automation, not being displaced by it. This is the emotional reversal Fraser's memo primed them for.
• Move fast. The 90-day payout means their financial runway ends on a specific, near date.

A founder who does this well can close a staff-level ex-Citi platform engineer in a two-week loop. A recruiter who leads with "exciting opportunity at a fast-growing startup" gets ignored, because the candidate has been fielding those messages for 90 days already.

What the pool looks like in numbers

Citi's headline figure is 20,000 roles eliminated globally against a September 2025 workforce of 227,000, or roughly 8.8% of the company. As of January 2026, Fraser said Citi had shed more than 10,000 roles, meaning the September 2026 tranches sit in the back half of the plan and the pipeline continues into 2027.

20,000

The strategic point for sourcing bank engineers: this is a repeatable playbook. American banks have been trimming jobs led by Wells Fargo and Goldman Sachs alongside Citi. Any founder who builds a workflow for the 9/30 Citi tranche can reuse it verbatim for the next Wells Fargo or Goldman WARN. NYC tech talent in September 2026 is dominated by this dynamic: the AI-labeled startups are hoarding attention while the traditional-finance modernization pool leaks quietly into the market.

Bottom line

Citigroup's layoffs in 2026 are not a story about a struggling bank. They are a story about a two-year modernization program winding down and releasing the engineers who built it. The Citi NYC WARN notice for September gives you three clean dates, one concentrated geography, and a candidate pool that has been on notice long enough to move quickly. The sourcers who filter out "bank" will miss it. The ones who read the memo and know that VP means IC will hire the whole top of the list.

FAQ

How many ex-Citi engineers will actually hit the NYC market on 9/10, 9/12, and 9/30?

Citi has not published a per-date headcount, and WARN filings often understate the true engineering share of the affected pool because they aggregate across job families. What is known: 69% of Citi's 2025-26 WARN filings land in New York State, roughly 20% of Refolk's Citi engineering sample sits in NYC Metro, and roughly 10,000 of the 20,000 planned cuts remained as of January 2026. Treat the three tranches as a few hundred technical roles combined, concentrated in platform, SRE, and application engineering, and size your outreach accordingly.

Are Citi engineers really working on modern stacks, or is this Java-and-mainframe territory?

Both, but the ones being cut skew modern. Citi's Transformation program is a $2.5 billion effort explicitly targeted at upgrading aging technology, and Fraser said in January 2026 that more than 80% of it is complete. The engineers being released in September are disproportionately the ones who spent the last two years shipping the modernization that replaced the legacy stack. The mainframe holdouts are not the ones on the WARN list.

What does a Citi "Vice President" title actually mean for sourcing?

At Citi and most large banks, Vice President is a mid-to-senior individual contributor title, roughly equivalent to a staff engineer at a tech company. It is not an executive role. If your search string filters on "Software Engineer" or "Senior Engineer" alone, you exclude the entire senior IC layer, which is exactly the profile a founder or engineering leader is trying to hire. Include AVP, VP, and SVP in your title targets and translate them internally before you reach out.

Why sequence outreach across the three dates instead of blasting once?

Because the three dates almost certainly represent three different Citi entities or business units, and the candidates on each list talk to each other. A single blast on 9/9 burns your first-message quality by 9/12 and 9/30, when a fraction of the later cohorts have already seen the earlier message forwarded internally. Staggered campaigns with different openers, keyed to each closure date and the 90-day payout endpoint, convert materially better than one wave of identical InMails.