Trase Just Raised $107M for 80 Seattle Hires. The Koneru Alumni Pool Is 9.
Trase's $107M seed targets 80 Seattle engineers. The AWS Bedrock Agentic alumni pool is 9. Here is the actual sourcing lane, with numbers.
On June 24, 2026, Trase announced a $107M seed led by ARCH Venture Partners and confirmed plans to grow its Seattle engineering hub from about 20 people to roughly 100. The two names at the top, Baskar Sridharan and Srirama Koneru, both came from AWS, and both point to a very narrow alumni graph. If you are a recruiter or founder trying to poach the same profile, the math is uglier than it looks.
Trase's $107M seed is really a bet on two AWS alums
Trase is a Red Cell Partners-incubated agentic AI startup building autonomous agents for healthcare, national security, and energy customers under HIPAA and SOC2 constraints. The $107M seed, announced June 24-25, 2026, was led by ARCH Venture Partners with Red Cell participating, and it more than triples the company's war chest as it scales from 55 employees to a Seattle-heavy 135-ish over the coming months.
The headline hires are the story:
- Baskar Sridharan, president. Ex-VP of AI/ML services and infrastructure at AWS. Before that, VP of engineering at Google Cloud. Before that, roughly 16 years at Microsoft architecting Azure storage.
- Srirama Koneru. Former GM of Bedrock Agentic AI Infrastructure and GenAI Services at AWS. Former senior director of engineering at Google and Salesforce.
- Grant Verstandig, CEO of Trase and founder/CEO of Red Cell Partners, the venture studio that incubated the company.
Sridharan is the recruiter magnet. Koneru is the sourcing lane. Verstandig is the check. The question every competing recruiter should be asking is which of those three actually maps to a real talent pool you can hunt.
The Koneru alumni pool is 9 people
In Refolk's index of professional profiles, exactly 9 people in the United States carry "Bedrock Agentic AI Infrastructure" language in their headlines. That is the literal Koneru org, the engineers who built the agentic layer of Bedrock rather than just called its APIs.
This is the number nobody publishes because nobody bothers to distinguish "worked on Bedrock" from "used Bedrock." The distinction matters. Trase is not hiring people who wrapped a foundation model with a retry loop. It is hiring people who built the retry loop, the routing plane, the tool-invocation layer, and the tenancy model underneath.
Broaden the lens to anyone with "AWS Bedrock" experience in an engineering title (Software Engineer, ML Engineer, Applied Scientist, AI Engineer) and the pool grows to 101 in the US. Only about 18 of those sit at Senior, Manager, or Director seniority. Seattle metro accounts for 9 of the 101, or roughly 9 percent of the national pool, tied for the top region.
Against Trase's 80-hire Seattle target, that is an 8.9x demand overhang from a single startup. Trase alone would need to absorb the entire local Bedrock alumni pool nine times over.
The actual sourcing math, in one table
Here is what the pool looks like when you segment it the way a recruiter would actually work it. All counts are from Refolk's index, US-only, as of the round announcement.
| Cohort | Count | What it means |
|---|---|---|
| "Bedrock Agentic AI Infrastructure" in headline | 9 | The literal Koneru direct alumni. Most in Seattle/Bellevue. |
| "AWS Bedrock" alumni across engineering titles | 101 | The pool most recruiters chase. Seattle leads. |
| Same pool at Senior/Manager/Director seniority | 18 | About 18% of the broader pool is leadership-adjacent. |
| Seattle metro share of the 101 | ~9 | Concentrated but not dominant. SF Bay and NYC hold ~2% each. |
| Trase Seattle hires vs. Seattle Bedrock alumni | 80 vs. 9 | 8.9x demand overhang from Trase alone. |
| Koneru direct alumni vs. total Trase hire target | 9 vs. 80 | A 100% capture rate covers 11% of the roles. |
The takeaway is not "Trase will fail to hire." Trase will hire. The takeaway is that "AWS Bedrock alumni" is not the sourcing lane. It is the press release. The real sourcing lane is much wider and much stranger, and I will get to it.
Why the AWS internal pivot is the actual talent unlock
The reason Trase can plausibly land senior Bedrock engineers right now is that AWS has quietly stopped being a good place to be one. Two dated facts drive this:
- February 5, 2026: AWS India laid off nearly 400 of about 450 Bedrock employees in Chennai and Bengaluru. An internal email from AWS VP Barry Cooks signaled a strategic shift favoring Amazon SageMaker as the primary platform for AI training and customization.
- July 15, 2026: Dave Brown, the AWS SVP who oversaw both Bedrock and SageMaker, announced his departure after nearly 19 years for a role outside the company. His direct reports are, as of this article, in transition.
Zoom out and the picture gets worse for AWS retention. Amazon has eliminated more than 16,000 positions in 2026, the single largest contributor to this year's tech layoff wave, with cuts concentrated in AWS professional services, Alexa AI, Prime Video, and Amazon Pharmacy. Internal Slack messages surfaced by reporters show the cuts reached engineers on core cloud services including Bedrock and Redshift.
Translation: Bedrock-specific expertise is now a devalued internal asset at AWS. The people who built the thing are watching leadership pivot the strategy underneath them. Trase's pitch writes itself.
Come build the Bedrock you thought you were building. Not a comp argument. A mission argument.
This is where a plain-English sourcing pass beats a boolean string. If you tell Refolk "AWS engineers who shipped agentic infrastructure on Bedrock and are based in Seattle or Bellevue," you get the nine. If you tell it "AWS engineers who were reorged out of Bedrock into SageMaker in the last 12 months," you get a different, larger, and arguably more poachable list. That is the exact gap Refolk closes for Trase AI hiring: describe the person, not the keyword combination.
The regulated-industry engineer does not exist yet
Trase's flagship customer is Duke University Health System's cardiology division, where its AI routing completed fax triage 7.1 times faster than manual processing, saved staffers 1,395 hours per month, and unlocked $285,450 in annual staff capacity. That is a HIPAA workload. Trase is also SOC2 compliant and supports on-prem and edge deployment for national-security customers.
The uncomfortable fact: in Refolk's index, the intersection of "agentic AI infrastructure" and "regulated healthcare compliance" across engineering titles returns essentially zero results. That profile does not exist in the market. Trase cannot hire it. Trase has to manufacture it, by pairing AWS agentic-infra alums with compliance veterans from a different lane entirely.
The four companies to raid for the compliance half of the pairing:
- Epic Systems (Verona, WI). Deep HIPAA muscle memory. Historically hard to poach; the current AI wave is loosening things.
- Palantir (Denver/DC/NYC). FedRAMP High and IL5 engineering culture. Overlaps with Trase's national-security customer set.
- Anduril (Costa Mesa/DC). Defense-grade software supply chain and on-prem deployment experience.
- Cerner / Oracle Health (Kansas City). Post-acquisition attrition has been generous. Regulated-industry AI engineers who want to leave already know they want to leave.
None of those companies show up when you search "Bedrock." That is the point. The Trase hiring plan is a pairing exercise, not a keyword search, and any recruiter working it as a keyword search will lose.
Seattle is right, but not for the reason it looks
Seattle is the correct bet for Trase's build-out, but the ~9 Bedrock-engineering alums in the metro are not why. The real reason is the adjacent pool: Amazon has cut roughly 30,000 corporate roles since October 2025, Microsoft's Azure AI org sits 20 minutes away in Redmond, and Google Cloud has a substantial Kirkland footprint. Seattle AI engineering talent is deep on Azure and GCP infrastructure engineers who have never touched Bedrock but have built exactly analogous systems.
Sridharan is the tell. His pedigree is Microsoft Azure storage, then Google Cloud engineering leadership, then AWS. He was not hired to recruit only from AWS. He was hired to recruit from all three hyperscalers. The likely composition of Trase's 80 Seattle hires:
- 8 to 12 from the direct AWS Bedrock and adjacent Bedrock-agentic orgs (Koneru's network).
- 20 to 30 from Azure AI infrastructure and Azure ML, pulled via Sridharan.
- 10 to 15 from Google Cloud Vertex AI and adjacent GCP AI infra, also via Sridharan.
- 15 to 20 from the broader post-layoff AWS pool who were not on Bedrock but built analogous distributed systems.
- 10 to 15 compliance and regulated-industry hires from Epic, Palantir, Anduril, Cerner.
If you are competing with Trase for the same profile, the useful move is to source against that composition, not against the Bedrock keyword. That is how you find the Seattle AI engineering talent that Trase is quietly locking down.
The 9-person Koneru graph is a defensive moat
For competitors like Anthropic, Cohere, Sierra, and Vercel's agent-platform team, the fact that Trase has already locked up the top of the Bedrock Agentic org means the next two hires from that 9-person pool become disproportionately valuable. This is a classic small-cohort scarcity dynamic in agentic AI startup hiring.
Expect three consequences over the next 90 days:
- Signing-bonus inflation on the narrow cohort. For the ~9 Koneru-direct alumni still in-market, signing bonuses in the $500K to $1M range are plausible. There simply are not enough people to negotiate downward.
- Second-order poaching accelerates. As Trase pulls Bedrock alumni into Seattle, the rest of the Bedrock org at AWS becomes easier to poach because the reference pattern is now visible. Peers watch peers leave. This is how a 9-person pool becomes a 30-person migration within two quarters.
- Regulated AI engineers become their own category. Trase's Duke Cardiology traction, combined with ARCH Venture's healthcare-first thesis, will legitimize "regulated AI engineer" as a compensation band separate from generalist ML engineers. Watch for it to show up in comp benchmarks by Q1 2027.
For recruiters at competing agentic-AI companies, the practical response is to build the map before Trase closes offers. That means indexing the AWS Bedrock alumni diaspora, the Azure AI and Vertex AI adjacent pool, and the compliance-heavy roster at Epic, Palantir, Anduril, and Cerner in one place. Doing that in LinkedIn's UI takes weeks. Asking Refolk in plain English takes an afternoon, which is the difference between reaching these engineers before Trase does and reading about their new roles on GeekWire.
FAQ
How many people actually built AWS Bedrock's agentic infrastructure?
In Refolk's index of professional profiles, 9 US engineers carry "Bedrock Agentic AI Infrastructure" language in their headlines. That is the literal Koneru-org alumni graph. A broader "AWS Bedrock" filter across engineering titles returns 101 nationally, but that pool includes people who used Bedrock rather than built it, which is a materially different profile for Trase's needs.
Is Trase's 80-hire Seattle target actually feasible?
Yes, but not from the Bedrock alumni pool alone. Seattle metro has roughly 9 Bedrock-engineering alums, and Trase's target creates an 8.9x demand overhang against that specific cohort. The plan only works if Trase leans heavily on Sridharan's Azure and Google Cloud graph, plus the broader post-layoff AWS engineer pool, plus compliance hires from Epic, Palantir, Anduril, and Cerner.
Why is now a good moment to poach AWS Bedrock engineers?
Two dated events: on February 5, 2026, AWS India laid off nearly 400 of 450 Bedrock employees and signaled an internal shift toward SageMaker, and on July 15, 2026, AWS SVP Dave Brown, who oversaw both Bedrock and SageMaker, announced his departure. Combined with more than 16,000 Amazon job cuts in 2026, Bedrock-specific engineers inside AWS are the most poachable they have ever been.
Where does the regulated-industry AI engineer come from if the profile does not exist?
Trase will manufacture it by pairing AWS agentic-infra alums with compliance veterans from Epic (HIPAA), Palantir (FedRAMP High, IL5), Anduril (defense on-prem), and Cerner/Oracle Health (post-acquisition attrition). No single company today ships engineers who have both agentic AI infrastructure and regulated healthcare compliance in the same resume, which is why competing recruiters have to source the pairing, not the keyword.