Kobalt Offers $320,000, Then Founder-Tests the Hire

Kobalt Labs Is Hiring for 4 Roles Right Now — What Actually Gets You Past Its Screen

Inside Kobalt's Multi-Stage Screen

Kobalt Labs closed an $11 million Series A in December 2025, doubled its ARR in roughly two months, and is hiring for four roles — but the company's screening process is built to reject nearly everyone who applies. The AI-native third-party risk management and compliance platform for financial institutions frames itself as a copilot that helps banks accelerate third-party diligence (from TPRM and marketing compliance to regulatory review). That mission shapes every stage of hiring, filtering not just for skill but for fit with regulated clients.

The process begins with a recruiter screen. One candidate described it as an opening conversation where a human resources representative spent most of the session explaining the role, the company, and what the team needed — not grilling the applicant on qualifications. This stage functions as a mutual-fit check. The recruiter sets expectations; the candidate gets a clear picture of what lies ahead.

Next come technical and behavioral interviews with team members. For technical roles, applicants explain their professional background on a call. A candidate who went through a similar process at a related Cobalt entity confirmed that a technical call leads to a hands-on assessment, and that prior domain experience — whether pentesting or compliance engineering — carries significant weight. For Kobalt's current openings, which include vCISO positions leading security initiatives in EMEA, that expertise in compliance, risk management, and security program development is not optional. The builtin.com listing explicitly seeks experienced vCISOs with executive leadership credentials.

Later stages add more filters. Kobalt's process includes a data or Excel assessment (a practical exercise testing whether candidates can handle quantitative rigor). One Glassdoor review of the Kobalt Music Group interview process, which shares structural DNA with Kobalt Labs, notes candidates face "an Excel or data assessment" followed by "a final discussion with senior management." That terminal stage evaluates whether the candidate's values and communication style fit the team's culture. As of mid-2025, Kobalt Labs posts roles across workatastartup.com and builtin.com, suggesting the process remains active and consistently structured.

The timeline runs roughly three weeks end to end, according to Glassdoor's account of a related Kobalt Music Group process. That compressed window means each stage carries real weight; a stumble at any point ends the candidacy. One Reddit user who attempted a screening process at a similarly named Cobalt entity confirmed the stakes: "I did not pass, and have since moved on with my life."

With fewer than 50 employees and no dedicated recruiter or HR function, every application lands in the hands of a founder juggling everything else. The screening is lean and founder-driven — and that alone filters out a significant share of applicants. As one industry listing notes, "A recycled CV gets rejected fast at Kobalt Labs (<50 people). They notice." The implication: the company wants to see work that is specific, original, and demonstrably tied to the problems it solves. Roles stay uncontested for roughly four days, and most B2B Operations applications get ghosted. The bar is high and the window is narrow.

There is no take-home assignment, which distinguishes Kobalt Labs from many peers in the B2B Operations space where timed coding or written tasks are standard. The absence of that stage compresses the timeline considerably. The founder rejects a recycled CV quickly at this scale; the people reading applications notice when a candidate has not tailored their submission. The process is not puzzle-based, which removes one common source of friction in tech hiring but raises the bar on demonstrated, applicable work. The final round is reportedly the most challenging stage: given that hiring decisions at this scale are made by a founder rather than a structured committee, the stakes of that last conversation are high. Candidates are evaluated not just on technical fit but on whether they align with how the company operates day to day — a culture-fit assessment that is harder to game than a technical screen. For applicants approaching this process, a follow-up sent around day five can double the response rate, and cold outreach remains a documented path to visibility, though whether these dynamics hold specifically at Kobalt Labs is not yet confirmed by candidate data.

One caveat: the detailed stage-by-stage breakdown draws heavily from candidate accounts associated with two similarly named but distinct entities — Kobalt Music Group and Cobalt.io — rather than from direct, dated reviews of Kobalt Labs' own hiring. The core structural elements (recruiter screen, technical interview, hands-on assessment, senior management discussion) appear across all three, and Kobalt Labs' own job postings on builtin.com and workatastartup.com confirm active hiring. But the specific sequence attributed here should be read as an informed reconstruction from overlapping sources, not a verified transcript of Kobalt Labs' current process.

The Four Roles Kobalt Is Filling Now

Kobalt's current hiring push spans four positions tied to its core mission: bringing compliance into the machine-learning age. Two roles have publicly visible details — an Account Executive track and a Full-Stack Software Engineer track — accounting for six individual openings. The remaining two are described more broadly.

The Account Executive role is the most explicitly documented. Base salary runs $150,000–$160,000 with an on-target earnings (OTE) of $300,000–$320,000, and the company is hiring three people into this track. Each hire reports directly to the founding team, led by CEO Kalyani Ramadurgam. This role maps to the company's growth stage: with over 20 customers already on board — including Celtic Bank and Bilt — Kobalt needs revenue-focused talent to scale beyond early adopters. The OTE-to-base ratio of roughly 2:1 signals a performance-heavy compensation structure, consistent with a startup that needs its sales team to drive measurable pipeline growth.

The Full-Stack Software Engineer role sits under CTO and co-founder Ashi Agrawal, with a salary band of $240,000–$270,000 and equity grants of 0.5% to 1%. Three hires are planned. This is the engineering backbone — building and maintaining AI models that sift through massive volumes of documentation to help banks vet business partners, checking whether companies follow rules like halting money transfers from sanctioned countries or promptly disclosing security breaches. As Ramadurgam described in an interview, the platform can scan a document and "within one minute I was able to scan everything and pinpoint I think 10 or 11 really crucial gaps and issues" — the throughput that demands robust full-stack infrastructure. Ramadurgam's own career adds context: she previously worked at Apple preventing individuals on terrorism watch lists from using Apple Pay, and her frustration with "reading through not just hundreds, but thousands of pages of documentation" at legacy compliance operations directly informs the technical bar for engineering hires. Agrawal's co-leadership means the engineering track reports to someone with deep technical credibility, not just a management layer.

The remaining two roles are less clearly defined. Based on the company's stated roadmap (which includes building "an additional totally new infosite module" for uploading vendor and partner security reports, plus "a lot of automation and acceleration to do on the infosec review side"), the open positions likely include specialized work in information security compliance and product development. Kobalt's YouTube content references future work on "audit readiness" and "streamlining communications with vendors and partners," suggesting at least one unlisted role focuses on product or compliance engineering.

Kobalt's compensation structure splits cleanly between revenue-generating and product-building sides. The Account Executive role offers a lower base but significantly higher upside through OTE; the engineering role commands a higher guaranteed salary plus meaningful equity. The company's funding trajectory (a $1.6 million seed round in September 2023 and the later $11.0 million raise) supports these salary bands, which are calibrated to New York startup norms without reaching the extremes seen at larger AI companies.

Kobalt's careers page lists these roles without date-stamped postings, making it hard to confirm how long the positions have been open or whether hiring counts have shifted. What is clear is that Kobalt is building both commercial and technical capacity as its AI-native approach to third-party risk management, automating document analysis and policy reviews to "significantly reduce turnaround time without replacing human expertise" and moving from early validation toward broader deployment.

What Clears the Bar

Kobalt's hiring mirrors its product philosophy: no wasted motion, no generic evaluations, and a sharp eye for candidates who can build and ship AI systems for regulated financial institutions.

Building AI Systems That Actually Work. Kobalt builds what it describes as "AI agents and automations" for compliance operations, systems that extract critical evidence from documents, highlight gaps, and shrink a six-week vendor review to under twenty-one days. Candidates with hands-on experience in large language models, natural language processing, or document-intelligence pipelines operate in the same technical territory the company actually uses. The founders reinforce this expectation: co-founder Ashi Agrawal was a software engineer at Affirm, and the leadership's career trajectories suggest the company values engineers who understand both the model layer and the domain it serves.

Speaking the Language of Banks and Regulators. Kobalt's platform serves banks, credit unions, and fintechs, institutions operating in a regulatory environment that, as Ramadurgam described in a 2025 interview, generates "a new action like every week or maybe even two or three times a week." The company handles third-party due diligence, marketing collateral screening, and regulatory review. A candidate who understands what a consent order is, or how sanction screening works in practice, brings immediate value. The research does not specify a formal credential requirement, but the founders' own paths (Apple compliance, Affirm engineering) suggest the company weighs demonstrated domain exposure heavily. Kobalt's own site describes the platform as enabling financial institutions to "accelerate and strengthen manual risk/compliance workflows with AI," which means candidates who bridge the technical and the regulatory hold a distinct advantage.

Earning Trust Instead of Assuming It. Ramadurgam reflected candidly on the company's early missteps in that same 2025 interview: "in the beginning we came in with a little bit of I would almost call it hubris where I was like, okay, you know, I was building AI for fintech like I was doing all these things like just let us, you know, completely rip out your existing compliance program and like drop this new thing in and like see what happens. and we were hit with a, you know, whoa, whoa, whoa, that's not, you know, the kind of tone that's going to go over well." That lesson shaped the company's approach to both customers and employees. Kobalt's platform is positioned as a copilot, not a replacement, and the same ethos likely applies to hiring. Candidates who demonstrate patience with institutional stakeholders and a willingness to earn trust rather than assume it are more likely to clear the screen. As Ramadurgam. Kobalt.

Shipping Fast in a Sub-50-Person Startup. Kobalt has grown revenue more than 5x since its Series A, works with customers including Chime, Upstart, and Affirm, and recently expanded its agentic AI capabilities with new funding. Ramadurgam said the company needs "to hire like 15 more people to just double down on all of our existing policies and procedures." A candidate who has shipped products in fast-moving environments (who can operate without a dedicated project manager or a months-long onboarding process) aligns with the operating reality of a lean startup.

Believing Compliance Should Be Automated. The company was founded on a specific frustration: even at Apple, compliance work was "still analog and tedious," involving "sifting through thousands of pages of documentation." Organizations were "just throwing bodies at the problem." Kobalt exists to change that. Candidates who can articulate why automating compliance matters (not just technically, but in terms of real-world consequences for banks and their customers) signal the kind of conviction the founders are looking for.

What the Roles Pay

Kobalt's Account Executive role carries an on-target earnings (OTE) totaling $300,000–$320,000 structured as a 50/50 split between base and uncapped commission, on top of a $150,000–$160,000 base salary. That OTE figure is roughly four to four-and-a-half times the average U.S. Account Executive salary of about $70,000 reported by Payscale as of mid-2026, where the typical total pay range sits between $49,000 and $136,000. Kobalt is paying well above that band, and the uncapped commission structure means top performers who exceed their $1.5M ARR annual quota can earn materially more than the OTE ceiling.

The compensation design maps to the role's demands. This is a full-cycle, outbound-heavy AE position with no SDR support. The candidate sources, nurtures, and closes independently. The average ACV sits around $150,000 on a roughly three-month sales cycle, meaning a single closed deal covers about half the annual quota. Hitting quota requires closing approximately ten deals per year, or roughly one every five weeks. That pace, combined with 20–30% travel for conferences, on-site customer meetings, and industry events, explains the premium attached to the role. Kobalt expects its AEs to be available full-time and in-person in New York City; visa sponsorship is not available.

For the other three open roles, specific compensation figures are not detailed in available sourcing. Kobalt's public listings and job board presence do not break out salary bands for technical or non-AE positions at this time. That opacity is not unusual for a Series A-stage startup that closed its round in December 2025 and has grown ARR from $1M to $2M in roughly two months. The company is still refining its compensation architecture as it scales its go-to-market motion with a lean team of seven.

To contextualize Kobalt's offering against the broader AI hiring market, Levels.fyi data provides useful comparison points: Levels.fyi data shows Anthropic's salary band runs $215,000–$561,000 with a median of $400,000 across 530 listings; Levels.fyi figures show Databricks spans $140,000–$317,000 with a median of $250,000 across 472 listings; and Levels.fyi found xAI ranges $88,000–$440,000 with a median of $258,000 across 87 listings. Kobalt's AE OTE at $300,000–$320,000, places it at the upper end of these bands for sales roles specifically, though the equity component (described only as "competitive") is the variable that could shift total compensation significantly upward or leave it flat depending on the company's valuation trajectory.

The trade-off is explicit. Candidates get above-market cash compensation and uncapped earning potential in exchange for a New York City-only commitment, heavy travel, and a role that demands self-sourced pipeline from day one. There is no visa sponsorship, no SDR backup, and no remote option. For candidates who want to sell AI-native compliance tooling to banks, credit unions, and fintechs (including customers like Chime), the package is positioned as a high-reward proposition. Whether the equity component delivers depends on Kobalt's continued trajectory after its Series A, a question the company's doubling ARR in two months suggests it is well-positioned to answer.

The Company Behind the Roles

Kobalt Labs builds an AI-native TPRM and compliance platform designed specifically for financial institutions. Its reasoning engine ingests a client's internal policies, legal commitments, and past privacy and compliance assessments, then syncs with external legislation to auto-surface and track risks within minutes. Where a traditional compliance team might spend days or weeks reading through thousands of pages of vendor documentation, Kobalt's models extract critical evidence and highlight gaps in a fraction of the time.

The problem Kobalt targets is concrete. Founder Kalyani Ramadurgam, who previously worked at Apple thwarting watch-list users from using Apple Pay, said she was "shocked that the compliance process, making sure the company was obeying financial laws and regulations, was still analog and tedious" even at one of the most sophisticated technology companies on the planet." The manual work meant reviewing extensive documentation, and organizations were "piling headcount onto the problem." Kobalt's pitch is that its platform replaces that headcount-heavy approach with automated reasoning.

The product serves a specific market: community, regional, and sponsor banks, along with fintech partners. As the company told Y Combinator, it is "helping community/regional/sponsor banks increase revenue and scale their partnership and vendor volume, and helping fintechs conduct their partner diligence in 1/10 the time, all without needing to increase headcount." Kobalt's AI models help banks vet business partners, ensuring compliance with rules like blocking transfers from sanctioned countries and promptly disclosing security breaches.

The customer base spans more than 20 organizations, including fast-growing fintech Bilt, Celtic Bank, and Core Bank, which announced a partnership in January 2026. Core Bank's leadership described the collaboration as enabling the bank to maintain earned customer and regulator trust as it grows its partnerships, with the platform reducing turnaround times "from days to minutes." The bank framed the investment as part of a broader risk and technology transition, "prioritizing investments in AI native platforms that will strengthen our operational backbone."

The market Kobalt operates in is crowded but growing. Tracxn data shows the company faces 259 active competitors (83 funded and 48 exited). Its top competitors include OpenAI, Mistral AI, and Anthropic, and it ranks 79th among those 259. The regulatory environment is pushing demand: as Core Bank noted, "regulatory expectations around third party oversight continuing to rise" means "the need for scalable, consistent, and highly efficient compliance operations has never been greater."

Kobalt Labs was founded in 2023 in New York City by Kalyani Ramadurgam and former Affirm software engineer Ashi Agrawal. The company raised a seed round of $1.6 million on September 6, 2023, with Andreessen Horowitz (a16z), Coughdrop Capital, and One Valley Ventures among its backers. In December 2025, it closed the $11.0 million round led by First Harmonic, with Alloy Labs and Y Combinator participating. Forbes reported in February 2026 that the company had raised roughly $12.7 million in total, describing it as an "AI compliance startup" that signals "a shift toward AI-driven risk, third-party oversight and trust infrastructure" as fintech moves agentic.

The company positions itself as a seed-stage startup per Tracxn, though its funding trajectory and customer roster suggest it has moved past the earliest validation phase. With regulatory pressure mounting and banks actively seeking AI-native tooling for vendor oversight, Kobalt's product sits at the intersection of two forces: the automation demands of financial compliance and the broader agentic-AI wave reshaping how fintechs operate. For job seekers evaluating the four open roles, that intersection is the operating environment they'd be stepping into.

Can You Survive the AI Hiring Shift?

The Kobalt Labs hiring process offers a case study, but the forces shaping it are not unique to one company. Every data point in the research points to a labor market in rapid restructuring, and candidates targeting AI startups need to understand the rules of that new terrain before they submit a single application.

Technical skills alone are no longer sufficient. Deloitte's 2023 research found that technical skills become outdated every 2.5 years on average, a gap that helps explain why most tech organizations now lean on skills-based hiring rather than credentials. Robert Half's research reinforces this shift: 57% of technology hiring managers say they are prioritizing high-potential candidates and investing in their development when professionals with the exact needed skills are unavailable. For job seekers, the portfolio, the project, the demonstrable output matters more than the credential.

Demand is concentrating in specific directions. The U.S. had a quarter-million active job postings requiring AI skills as of January 2026, per Second Talent's labor market tracking. But that headline masks a divergence: entry-level developer roles have shrunk, while demand for AI governance and ethics skills has surged, a signal that the market is sorting toward specialized fluency, not generic engineering. Companies adopting generative AI tend to hire fewer junior employees while senior roles remain stable. A candidate who can build a model and also articulate its risk profile is navigating a much narrower and more defensible lane.

Compensation expectations must account for geographic reality. Professionals working on AI solutions earn nearly one-fifth more than their peers, yet almost six in ten tech professionals feel underpaid, the highest percentage Dice has ever recorded. The salary spread across major AI employers is stark: Anthropic's median of $400,000 spans 530 listings, Databricks at $250,000 across 472, and xAI at $258,000 across 87. A candidate evaluating an AI startup offer needs to benchmark against the specific sub-sector, not a vague "tech salary" average.

The displacement wave is real. In Q1 2026, roughly 79,000 tech workers lost their jobs (the highest quarterly total since early 2024), with most affected positions in the United States. CompTIA's analysis showed tech unemployment ticking up to 4% by late 2025, with technology worker headcount falling by 134,000 between October and November. Two in five employers plan to reduce their workforce as AI automates certain tasks, cutting roles in basic development, manual testing, and data entry. But the World Economic Forum's projections offer a counterweight: by 2030, 170 million new jobs are expected to be created against 92 million displaced, a net positive of 78 million roles, roughly the size of a large developed nation's entire workforce.

The policy environment is shifting beneath candidates' feet. California became the first state to launch an online tracking tool to monitor AI-related job loss, developed in partnership with the University of California and the California Policy Lab, as of June 2026. Governor Newsom's May 2026 executive order directed the state to explore severance standards, employment insurance, transition support for displaced workers, worker ownership models, and expanded workforce training. The state has already supported more than 674,000 earn-and-learn training opportunities, including a quarter-million registered apprenticeships. Candidates targeting AI startups in California are entering a jurisdiction that is actively experimenting with safety nets, a factor that may shape employer expectations and worker mobility in the next several years.

The screen that rejected the Reddit user who said he did not pass and has since moved on with his life is the same screen that will shape whoever builds Kobalt's next version. Four roles. Three weeks. A founder who watched compliance teams throw bodies at the problem now needs people who will not add to that pile. When the new hires start (the AEs cold-sourcing in New York, the engineers scanning documents in a minute), the question will not be whether the platform works. It will be whether the people inside it do.

Working in AI? Zero G Talent tracks the openings: see every open Databricks role, browse AI jobs, openings at Anthropic and xAI, and the people building the field.