Corgi Insurance adds 55 roles, senior software engineer pays up to $275k

Fifty-Five Roles and Counting

Zero G Talent's board data shows 55 salaried roles open at Corgi Insurance as of this month, nine added in the past week. The salary band runs from $60,000 to $270,000, median $150,000. That volume puts Corgi well past the experimental phase — this is a scaling operation with the capital to back it. The company's $4 billion valuation, confirmed in a Silicon Valley Investclub interview, reflects the same conviction: owning the full stack requires headcount in every layer, not just software.

Zero G Talent's figures put the Senior Software Engineer top at $275,000.

The spread crosses engineering, distribution, capital markets, and brand — evidence the "full stack" claim extends to the org chart. Corgi describes itself as an "AI-native full-stack insurance carrier" rebuilding commercial insurance from the ground up. That architecture demands engineers who can ship production-grade systems across policy administration, quoting, claims, and data pipelines, often simultaneously. The founder has said building a licensed carrier from scratch is "one of the hardest things a startup can do," requiring control of the entire stack: underwriting, reinsurance, claims adjustment, and a TPA handled in-house with minimal outsourcing. The interview put the target bluntly — anything under $100 billion in eventual value counts as failure. The 55 roles are the down payment on that trajectory.

What distinguishes this surge from a generic headcount push is the AI-native premise. The carrier was designed from day one to automate quoting, underwriting, and claims decisions that legacy carriers still route through manual review. Every engineering role on the board connects to that automation layer. Every distribution role connects to a product that can be priced and bound without human handoffs. The hiring map mirrors the architecture.

Inside the Screen: Code Over Credentials

The board's open roles cluster in engineering, sales leadership, and operations, heavy on senior individual contributors and domain-specific engineering. That distribution signals a filter built for technical depth in insurance workflows, not generalist pedigree. The research does not publish Corgi's internal rubric or interview scorecards. What the board data and founder commentary jointly reveal is a de facto filter: the company hires for the specific hard problems of a full-stack carrier (data ingestion from brokers, actuarial model deployment, claims automation, compliance tooling) and screens for evidence that a candidate has already solved adjacent versions of those problems.

The Pedigree Trap

The shift away from pedigree-based hiring in insurtech isn't a philosophical experiment — it's a response to a broken funnel. Criteria Corp reports 67% of HR leaders say AI-generated resumes are slowing their process, and 75% of resumes never clear ATS filters. When the signal-to-noise ratio collapses, throwing out the resume entirely starts to look like sanity. But the correction carries its own distortion.

AI screening tools are the mechanism enabling verification at scale. Job assessment and screening rank as the top ROI area for AI adoption in talent acquisition, with job matching second, per an iCIMS fireside chat with Tim Sackett. The economics are blunt: automated screening reallocates recruiter hours from resume triage to deep technical conversation, changing the signal-to-noise ratio of the entire pipeline. Candidate demographics reinforce the shift. Roughly 63 percent of tracked applicants fall in the Gen Z and millennial cohorts, and hiring managers report this generation prefers to showcase skills through assessments rather than interviews, per the same iCIMS discussion.

Bias reduction provides a second-order justification. Funnel analytics catch drift in real time: "Oh we get through some interviews now we're 70-30. We get to offers now we're 80-20. Like where are we? How is AI going? Hey, wait a minute. We're starting to slide to this bias, you know, and let's figure out where this is before it becomes a real problem." Data replaces accusation: "now going to have the data, you know, to be able to prove it. So that way it's not so accusatory. It's just going, 'hey, Mr. Miss manager, here's here's some data points.'"

Corgi's 55-role expansion, spanning enterprise sales, full-stack engineering, fund operations, and trucking insurance specialization, mirrors this industry-wide recalibration. The roles demand domain fluency that no transcript conveys: understanding how an ETF wholesaler thinks, how trucking claims adjusters document loss, how fund administrators reconcile NAVs. Those competencies are earned in production, not lecture halls.

Show, Don't Tell: The New Application

Public data on candidate behavior specific to Corgi's pipeline is thin; no surveys, forum analyses, or hiring-manager interviews detail how applicants are reshaping their preparation for this insurtech in particular. What exists are the job postings themselves and the broader pattern: when a company with 55 open roles signals it screens for demonstrable technical execution over academic background, candidates who want those roles adjust accordingly.

The salary bands confirm these are not entry-level experiments. They are production seats. In the wider applied-AI hiring market, engineers increasingly replace generic "machine learning coursework" lines with links to repositories that demonstrate model serving, not just model training. Data scientists pivot from leaderboard scores to end-to-end notebooks that ingest messy policy PDFs, extract structured fields with an LLM, validate against a schema, and write to a staging table, because that is the actual work inside an insurtech automation stack.

Open-source contributions carry weight when they map to a company's stack. A pull request that adds a retry policy to a Python HTTP client used in production, or a fix for a connection-pool leak in an async SQLAlchemy integration, reads louder than a master's thesis on attention mechanisms. Candidates who lack commit access to relevant projects create their own: a small library that normalizes carrier-specific JSON schemas into a common claim format, published to PyPI with CI/CD and semantic versioning. The artifact is verifiable; the interview conversation starts from code the candidate owns.

Portfolio sites shift from credential showcases to technical write-ups. A candidate for a senior insurance engineering role might publish a postmortem of a rate-calculation service they rewrote: the legacy interface, the shadow-traffic testing strategy, the latency reduction, the rollback plan that saved a deployment. That postmortem is a proxy for the systems thinking interview loops probe — failure domains, observability, migration risk. It also answers the behavioral question "tell me about a hard technical decision" before the interviewer asks it.

For non-engineering roles, the adaptation is narrower but similar. Sales candidates prepare demo recordings where they walk a prospect through an API integration flow, not a slide deck. They study public API docs or reverse-engineer the data model from product marketing pages, then rehearse the technical objection-handling a solutions engineer would face.

The gap between this industry-wide adaptation and Corgi-specific evidence is real. No data confirms how many applicants have actually shipped these artifacts for Corgi's roles, nor what the pass-through rate is from portfolio review to onsite. The company's screening rubric is not public. But the incentive structure is: 55 roles, high bands, skills-first framing. Candidates who treat the application as a credential submission are betting against the stated filter. The ones who treat it as a technical audition, shipping code, writing postmortems, and demonstrating domain-adjacent fluency, are aligning with the only signal the company has said it trusts.

When the Filter Becomes the Strategy

A screening process built entirely on demonstrable execution, including shipped models, open-source contributions, and production-grade code, systematically filters for people who have already had the chance to build. That chance is not evenly distributed. Candidates from elite labs, well-funded research groups, or companies with mature ML infrastructure accumulate deployable artifacts as a byproduct of their day jobs. Candidates from resource-constrained academic programs, smaller firms, or non-traditional pathways often have deeper theoretical grounding but fewer shipped systems to show. A portfolio requirement that rewards the former while ignoring the latter doesn't measure ability; it measures access.

The Criteria Corp framework emphasizes "talent signals" that reveal "what people can do, how they work, and where they can grow" — validated by 20-plus years of I/O psychology and 80 million assessments. But even scientifically validated signals reflect the environments where they were calibrated. If the validation corpus overrepresents candidates from certain hiring pipelines, the resulting benchmarks inherit that bias. ISO 42001 certification for AI management systems governs process transparency and risk controls; it does not guarantee the underlying training distributions are representative of the full talent pool Corgi might want to reach.

There's also the question of what "technical execution" means in early-stage insurtech. Corgi's open roles suggest a company building across multiple specialized domains. A candidate who has deployed fraud-detection models at a major carrier brings different evidence than one who has published on causal inference for policy pricing but hasn't shipped to production. The latter might actually be more valuable for a novel problem space where existing playbooks don't apply. A screen that indexes only on prior deployment risks optimizing for the last war.

The board data shows Corgi adding nine roles in seven days, with salary bands ranging from $75,000 to $300,000. That velocity pressures hiring teams to rely on legible, fast-to-evaluate signals such as GitHub profiles, leaderboard rankings, and prior insurtech tenure. But the roles themselves (VP of ETF Wholesales, Head of Fund Administration, Senior Software Engineer for trucking insurance) demand domain fluency that no code repository demonstrates. Over-indexing on technical artifacts can miss the regulatory intuition, distribution knowledge, or actuarial judgment that separates a working prototype from a compliant product.

None of this argues for returning to pedigree filters. The resume is broken; Criteria's data makes that clear. But the alternative isn't a binary choice between credentials and code. It's a more expensive, slower process: structured work-sample evaluations calibrated to the actual problems Corgi faces, paired with interviewers trained to assess reasoning about unfamiliar domains. That process doesn't scale as cleanly as an automated screen. It also doesn't produce the 50% time-to-hire improvement Criteria advertises in its case studies. For a company adding 55 roles in a surge, the temptation to automate the filter is real. The risk is that the filter becomes the strategy.

Zero G Talent's board data reported Corgi adding nine roles in seven days, with salary bands ranging from $75,000 to $300,000. The next nine will tell whether Corgi builds a filter that finds the builders it needs — or one that only finds the builders who already had the chance to prove it.

Working in frontier tech? Zero G Talent tracks the openings: see every open Corgi Insurance role, browse frontier tech jobs, the companies hiring, and the people building the field.