Poach Ex-Amazon AI Engineers Before Swami's Boomerang Email Lands

Amazon's AI agent org, run by AWS VP Swami Sivasubramanian, is emailing laid-off ex-employees with expedited AI/ML interview loops under a program internally labelled "Swami's Boomerang Reengagement Initiative." AWS Finance is running parallel outreach, and Amazon spokesperson Haley Silva confirmed the practice on record. If you recruit for any AI competitor, Amazon just published its own poach list. Move before the boomerang emails land.

What Amazon just told the market

Amazon publicly named the exact segment of its alumni it wants back most: ex-Amazon AI engineers and AWS talent with ML skills. That naming is the sourcing signal.

The mechanics of the program are worth reading carefully. Recruiters have told former staff that "a lot has changed since you were here" and highlighted new work in AI and machine learning. Some ex-employees have been offered an expedited interview process. AWS Finance recruiters are also asking whether Amazon's return-to-office policy contributed to departures, which tells you the internal team already knows RTO is the wedge competitors will use.

Fast-track loops do not exist unless internal demand is desperate and validated. Amazon has effectively pre-qualified the candidates, told you which teams to source, and admitted its own weakness (comp bands and RTO) in the same breath.

The pool: 30,000 cuts, and where the AI/ML value sits

Roughly 30,000 corporate roles are in play, but the AI/ML value is concentrated in two sub-pools: Alexa AI and AWS Professional Services. Everything else is noise for an AI/ML recruiter outreach list.

The layoff arithmetic, from public reporting:

• Around 16,000 layoffs in January 2026, plus roughly 14,000 in late 2025, for a total near 30,000 corporate roles cut.
• Washington state alone accounted for roughly 2,600 layoffs across Seattle, Bellevue, and Redmond, with more than half in product and engineering.
• Alexa AI lost "several hundred" positions in Sunnyvale and Seattle as Amazon pivoted to its LLM-native Rufus product.
• AWS Professional Services cut technical account managers and solutions architects.
• Amazon Pharmacy, Prime Video, and Amazon MGM Studios also cut, but those are less relevant to an AI/ML shortlist.

30,000

Alexa AI is the highest-signal pool because Rufus displacement means these engineers have real LLM and agent experience Amazon itself now wants back. AWS Pro Serve is the second pool for a different reason: solutions architects who know how enterprises actually deploy AI. Most AI startups have zero of that profile on staff. Different buyer, same source list.

Where these engineers already are

In a Refolk index sample of 20 US engineers with AWS plus ML skills and Amazon tenure, 7 of the top 10 still list Amazon as their current or most recent employer. The pool is currently captive, and that is exactly why the churn window matters.

The number that matters most: 7 of the top 10. This pool has not scattered yet. It is contactable, and about to receive an Amazon recruiter email. The competitor with a ranked shortlist in the next two weeks wins.

That is the exact gap Refolk closes for AI/ML recruiter outreach: describe "engineers who left Amazon in the last 60 days with AWS and LLM experience, based in Seattle or the Bay Area" in plain English and get the ranked list back, cross-referenced across GitHub, LinkedIn, and the open web.

Why Amazon's own comp rules are a moat for poachers

Amazon's boomerang policy pegs returning employees to their old salary band if they come back too quickly, and that arithmetic is a gift to competitors. A fresh SDE5 offer at Amazon was hitting $325k with 100% band penetration last summer, per Blind. A boomerang returning inside the cooldown window could be capped at their prior $220-230k.

That is a 30-50% total comp gap you can close without a bidding war. You are not fighting Amazon's recruiters. You are fighting a comp policy their own recruiters are quietly frustrated with.

The rehire cooldown itself, per widely reported Blind threads:

• Left in good standing: theoretically eligible the next day.
• Standard departure: 30 to 180 day cooldown depending on reason.
• Left while on PIP or pivot: indefinitely ineligible.

Laid-off-in-good-standing ex-Amazonians are the exact pool Amazon can move on fastest through Swami's initiative. They are also the pool you can move on fastest, because they cleared Amazon's bar recently and their references are live.

Amazon named the segment, tagged the resumes, and put a cooldown on their own comp. You just have to arrive first.

The window is 1 to 3 weeks, not months

Public reporting from golayoffs.com suggests laid-off Amazon AI engineers are landing in new roles within one to three weeks, which means "open to work" on LinkedIn is already a lagging indicator. The signal to source on is "left Amazon in the last 60 days," not "actively searching."

Yogesh Verma, a former AWS engineer laid off in January 2026, joined an AI marketing company. That landing pattern is typical, not exceptional. By the time a candidate updates their headline to "Open to Work," they have already taken three coffee chats and are close to signing.

The practical implication for sourcing:

1. Do not filter on "Open to Work." Filter on employment gap start date.
2. Cross-reference GitHub commit history for the specific weeks after their Amazon end date. Engineers who ship personal projects in week one are the strongest signal.
3. Pull the Alexa AI and AWS Pro Serve org charts from before the cuts, then diff against current LinkedIn.
4. Weight Seattle, Bellevue, and Redmond first. Sunnyvale second for Alexa AI specifically.

The two sub-pools worth building separate lists for

The Alexa AI pool and the AWS Pro Serve pool serve different buyers, and mixing them into one search wastes both. Build two lists.

Alexa AI (Sunnyvale, Seattle)

These are the engineers Rufus displaced. Rufus is Amazon's LLM-native replacement for the pre-LLM Alexa stack, so the people who got cut are exactly the ones who saw an in-house LLM migration up close. That is rare experience. Buyers: AI startups building voice agents, any consumer LLM product, applied teams at frontier labs.

AWS Professional Services

Technical account managers and solutions architects who deployed AI workloads into Fortune 500 accounts. They know Bedrock, SageMaker, and the actual enterprise procurement dance. Buyers: enterprise AI startups, NVIDIA's cloud teams, any GTM-heavy AI company that needs field engineers who can hold a CIO's attention.

The outreach: three angles Amazon just handed you

Amazon's own recruiter emails wrote your outreach for you. Three angles are now legitimized by Amazon's internal team asking the same questions.

1. The RTO wedge

AWS Finance recruiters are explicitly asking former staff whether return-to-office drove them out. That is a signed confession from Amazon's own team that RTO is bleeding talent. If you recruit for a remote or hybrid competitor, lead with location flexibility in the first line.

2. The comp gap

You do not have to name numbers. You just have to say "market rate for your level, not your 2023 band." Any ex-Amazonian who has read Blind knows exactly what that means.

3. The regret signal

Survey data cited by IBTimes suggests 55% of executives now regret replacing human workers with AI, and nearly half of companies that replaced staff with AI have experienced a return to human labour. This is a legitimate line for candidates weighing whether AI itself is a stable career bet. The answer for them is: yes, at companies where AI is the product, not the cost-cutter.

How to build the list this week

Six moves, in order, to have a working shortlist before Amazon's next boomerang email batch goes out.

1. Pull a list of every engineer whose LinkedIn shows Amazon as most recent employer with an end date in the last 90 days, filtered to AWS, Alexa, or AI/ML org tags.
2. Cross-reference against GitHub for commits after their Amazon end date. Personal repos with LangChain, LlamaIndex, or agent frameworks are strong signals.
3. Segment by sub-pool: Alexa AI vs AWS Pro Serve vs AWS Finance vs everything else. Do not mix.
4. Geo-weight: Seattle, Bellevue, Redmond first. Sunnyvale for Alexa AI. Then NYC and Bay Area.
5. Score by prior-art destination fit. If your role is at NVIDIA, weight profiles that have peers already at NVIDIA. Same for Meta, Apple, Salesforce, Instacart.
6. Send outreach in week one, not week four. The 1 to 3 week landing window is not a suggestion.

Google already ran this play. Amazon is late.

Google turned boomerang hiring into a documented AI recruiting engine in 2025, when about 20% of the software engineers it hired to work on AI were boomerang employees, per CNBC. Salesforce did it in 2023, encouraging former employees to return just months after cutting roughly 8,000 jobs. Amazon is copying a strategy that already worked twice, and AWS CEO Matt Garman has said Amazon plans to hire roughly 11,000 AI-focused engineers in 2026.

The competitors who beat Google and Salesforce to their own boomerang pools got the discount. The competitors who beat Amazon to Swami's list will get the same discount, on a pool Amazon itself has just certified as its top rehire priority. That closes the moment the first cohort of boomerang emails lands.

FAQ

Is "Swami's Boomerang Reengagement Initiative" a real, named program?

Yes, according to reports including IBTimes and StartupTalky, the internal effort has been labelled after AWS VP Swami Sivasubramanian, who oversees Amazon's AI agent organisation. Amazon spokesperson Haley Silva confirmed on record that boomerang hiring is a regular practice at the company, though she framed it as not exclusive to AI or cloud. The named-program detail is the sourcing signal: Amazon is treating this pool as a priority worthy of a fast-track loop.

Which ex-Amazon teams should I prioritize for AI/ML sourcing?

Two sub-pools deliver the most signal. Alexa AI in Sunnyvale and Seattle, where Rufus displacement freed engineers with genuine in-house LLM migration experience, and AWS Professional Services, where solutions architects and technical account managers know how enterprises actually deploy AI. Everything else in the 30,000-role cut (Prime Video, MGM Studios, Amazon Pharmacy) is largely noise for an AI/ML recruiter outreach list.

How fast do I need to move on sourcing laid off AWS engineers?

Faster than you think. Public reporting suggests many AI engineers laid off from Amazon are landing in new roles within one to three weeks, which means "Open to Work" on LinkedIn is a lagging signal. Source on employment gap start date and GitHub activity in the weeks after departure. If you wait for Amazon's boomerang email to trigger a candidate response, you are already competing against Amazon's expedited loop.

What is the strongest outreach angle when I poach Amazon alumni right now?

Location flexibility. AWS Finance recruiters are themselves asking ex-employees whether return-to-office drove them out, which is a signed admission that RTO is the wedge. Lead with remote or hybrid if you have it. Follow with market-rate comp, since Amazon's boomerang policy caps returners at old salary bands and competitors can beat Amazon by 30 to 50% TC without a bidding war. Close with prior-art proof: engineers with the same background have already landed at NVIDIA, Meta, Apple, Salesforce, and Instacart.