FINNY AI’s 11× Growth Is Hiring Two Engineers Who Know Wealth Tech

The Screening Gauntlet

FINNY AI's hiring push arrives with the velocity of a company that just convinced Venrock, Y Combinator, and Altruist founder Jason Wenk to back a $17 million Series A at a $150 million valuation in March 2026. The headcount target: two open engineering roles atop a 35-person team founded in 2024. The company's product logic reveals what matters. FINNY builds AI that scores prospects on 1.8 billion intent signals, updated daily, then ranks every match with an F-Score predicting conversion likelihood. The same probabilistic thinking shapes how the founders evaluate talent. Theodore Janson, Eden Ovadia, and Victoria Toli launched from Y Combinator's Summer 2024 batch with a thesis: wealth management's organic growth problem is a data problem, and the solution is multi-modal outreach triggered by behavioral evidence.

FINNY sells to registered investment advisors (400 firms paying $6,000 annually as of early 2026, a fraction of the 100,000-firm addressable market). Advisors waste an average of 58 hours chasing unqualified prospects before converting one client, per the company's data. The platform's newest features — Intent Search, Prospect Enrichment, AI Voicemails — all solve the same workflow fracture: advisors have rich client data but no signal on when to act. The stack ingests marketing data, household-level life events, and real-time search behavior to surface "money-in-motion" moments: job changes, inheritances, home purchases, retirements. That pipeline demands engineers who've built event-driven architectures, not just LLM wrappers.

No public source confirms the exact interview sequence (take-home, live coding, system design, founder conversation) or pass rates at each stage. FINNY doesn't publish a careers blog. Its Y Combinator profile lists two engineering openings; the company's own site and LinkedIn point to a waitlist, not a jobs page. What's clear is that the product reflects a high-signal, multi-modal approach weighted toward demonstrated judgment over credential density.

Two Engineering Roles, One Product Logic

FINNY AI's Y Combinator listing shows a 35-person team in New York City hiring for two engineering roles. The startup, founded in 2024 by Janson, Ovadia, and Toli, positions itself as an "AI growth officer for financial advisors." Its core engine scans millions of households for life-event signals — new home purchases, job changes, inheritances, retirements — and ranks each match with a proprietary F-Score measuring conversion likelihood. That description implies a stack heavy on large-scale data ingestion, real-time scoring infrastructure, and feature engineering that turns noisy consumer signals into advisor-actionable leads.

First-party board data for comparable AI-native companies (Anthropic, Databricks) shows senior engineering bands running $350k–$850k at the top end, but FINNY's Series A stage and 35-person headcount suggest bands closer to $180k–$300k base with meaningful equity, a gap candidates should clarify before investing interview cycles. The company's self-description ("from hello to close… guiding every step") hints at a product surface area spanning advisor-facing dashboards, CRM integrations, and the scoring API itself. Candidates who map their experience to the full funnel (ingestion → feature store → model → advisor UI) will speak the language FINNY's founders use on their own homepage.

What Gets You Past the Screen

FINNY AI builds the very class of tool that is reshaping how candidates get filtered — AI that qualifies prospects at scale. Research on AI-first screens is consistent: candidates who treat an automated interview as a performance (rehearsed narratives, strategic pauses, "storytelling" arcs) tend to fail. Large language models driving these screens predict the next token; they do not infer intent. When a candidate pauses to think, the model often interprets silence as completion and advances to the next prompt, cutting off the answer. A recent survey found nearly 40 percent of U.S. candidates have withdrawn from a process because it included an AI interview, and 12 percent said they would refuse outright. The drop-off is real, but the fix is not to opt out. A Chicago-based recruiter with a decade of experience put it bluntly: "Refusing to use it will put you behind them, even if they are using it poorly."

The tactics that work are mechanical. Start every answer with an explicit yes or no. Enunciate. When the bot asks a follow-up that embeds your own gibberish (a known failure mode where the model parrots noise back as a question), interrupt and ask for a restatement. The system allows that. Candidates who do so signal they understand the tool's limits and can steer it.

Domain fluency is a filter. FINNY AI sells to registered investment advisors and, increasingly, bank wealth-management teams. A candidate who can speak to that workflow (who knows what a Form ADV discloses, who understands the difference between a wirehouse and an independent RIA) reduces onboarding cost. The Y Combinator page notes the team has grown to 35; the TechCrunch profile from December 2024 put headcount at seven. Either way, the engineering-to-sales ratio is thin. Every hire must carry product context from day one.

Technical depth matters, but not in the abstract. FINNY AI's stack ingests public filings, property records, and professional-network signals to surface life-event triggers: a $5 million property purchase near Jackson Hole, a Form D filing, a trust amendment. Candidates who have built entity-resolution pipelines, worked with messy financial data, or shipped retrieval-augmented generation systems for regulated domains advance faster than generalist LLM wrappers. The seed round — $4.2 million co-led by Maple VC and HNVR — was explicitly earmarked for engineering expansion. The company's LinkedIn posts highlight a "Pay As You Grow" outcome-based pricing model for LPL advisors, which implies the team needs people who can instrument usage, measure attributable revenue, and iterate pricing logic in production.

Communication style is a differentiator. Candidates who communicate "in a slightly different way" (spelling out transitions, naming the framework before the example, labeling assumptions) succeed because the model has no human interlocutor to fill gaps. FINNY AI's CEO, Eden Ovadia, has said the product is a complement to traditional outreach, not a replacement. In practice, a cover letter or screen answer that references a specific FINNY AI feature (the waitlist of 250 vetted firms, the Morningstar Fintech Showcase win, the 11× month-over-month revenue growth in the first quarter) and ties it to a concrete problem the candidate has solved will stand out.

Inside the Company's First Eighteen Months

Finny entered the wealth-tech arena in late 2023 when Eden Ovadia co-founded the AI client-prospecting startup. Ovadia, who serves as CEO, has spent the company's first eighteen months pushing back on the industry's default skepticism. Registered investment advisors managing ultra-high-net-worth portfolios have long treated AI prospecting tools as either snake oil or a compliance headache waiting to happen. Ovadia's counter-argument is structural: she positions Finny not as a replacement for the referral networks that drive 40% of organic growth at firms like AlTi Tiedemann Global, but as a precision layer that makes those networks more efficient.

The product's core use case is narrow by design. Advisors use Finny to identify prospects around specific life events (someone who just sold a business, inherited a fortune, or bought a $5 million-plus property near Jackson Hole) then promote exclusive events tailored to that cohort. A Miami Heat suite invitation targeted at real-estate principals is the canonical example Ovadia cites. The platform also monitors existing clients for dissatisfaction signals, such as sudden spikes in online investment-advice searches. This dual motion (acquisition and retention) mirrors the workflow RIAs already run manually, but at a scale that would require armies of analysts to replicate.

Market timing explains why Finny is hiring now. In January 2026, Pathstone CEO Matthew Fleissig disclosed that his $182 billion RIA had fielded five inbound inquiries from clients worth at least $100 million each — all originating from AI search engines like Gemini and ChatGPT — in a two-week span. That signal, anecdotal but specific, shifted the conversation from "whether" to "how fast." AlTi's head of growth, Andrew Douglass, acknowledged the firm's 25-to-30-client annual target ($1.5–2 billion in new assets) and admitted openness to "a better mousetrap." A growth executive at another national RIA told Inside Wealth he had tested at least 20 AI prospecting demos in six months and found most built on the same five major LLMs with undifferentiated data overlays. Finny's differentiation bet is proprietary data pipelines and event-triggered intent signals rather than repackaged public records.

The competitive pressure is asymmetric. Incumbents can build internal tooling ("cents on the dollar," per the anonymous executive) but few have the product velocity to iterate weekly. Suno's product and tech teams shipped a screenshot-to-song feature in about a week, per Chief Product Officer Jack Brody, illustrating the cadence AI-native teams now expect. Finny's open roles map directly to that velocity requirement: engineering talent to harden the data layer. Ovadia's media engagement (on-the-record interviews with CNBC in January 2026) also functions as recruiting signal. Founders who articulate a clear thesis in public attract candidates who want to execute against it rather than explore. The roles are open because the company has moved past proof-of-concept into a market that just signaled readiness, and the window to compound that lead is measured in quarters, not years.

How Candidates Can Win

Greenhouse data shows the average corporate opening attracted 244 applications in 2025, more than double the 2022 figure, and Indeed reports the average time-to-offer has stretched to 47 days. In that environment, a generic resume is a losing ticket. Ohio University's career researchers found that "one of the biggest mistakes job seekers make is submitting the same generic resume for every position," and that formatting simplicity (clean headers, standard fonts, no columns or graphics) keeps AI parsers from dropping critical lines. Candidates should treat every application as a bespoke document: map each bullet to a requirement in the posting, quantify impact with metrics the screen can read (dollars saved, latency reduced, model accuracy lifted), and mirror the exact terminology the job description uses for tools, frameworks, and methodologies.

AI-assisted preparation is now table stakes, but the research draws a sharp line between using AI as a tool and outsourcing judgment to it. Ohio's sources emphasize that employers "want candidates who know how to use these tools effectively while still thinking critically and making strong decisions on their own." Practically, that means feeding the job description and your resume into a large language model and asking, "Here's the job, here's my resume, and here's the company: how can I optimize my resume for this role?" Then edit the output, verify every claim, and rehearse the narrative so you can defend it in a live interview. The same models can generate mock interview questions scoped to the company's stack, and candidates who practice aloud with timestamped feedback outperform those who only read prepared answers.

Networking remains the highest-leverage activity that no algorithm replicates. Nexford's transition framework urges job seekers to "seek out professionals who are already in roles you target; ask about how AI is impacting their work" and to "join online communities focused on AI in your domain." For FINNY AI specifically, that means identifying current engineers on LinkedIn, requesting 15-minute informational conversations about the team's evaluation philosophy, and feeding those insights back into interview stories. Penn State's Alumni Career Services builds its free Job Search Accelerator around exactly this loop: "how to tap into the hidden job market through relationships and networking" and "how to organize a search for maximum impact." The Kentucky Job Club's fall schedule dedicates a session to "AI and Your Job Search: Smarter Strategies for Today's Market" and another to "Creating a Resume That Gets Results: Standing Out in an AI-Driven Hiring World"; both free, public, and designed for the same bottleneck candidates face.

Skill gaps should be closed with visible artifacts, not just course certificates. Nexford recommends building "a mini-project: identify a real or simulated problem in your field where AI or automation could be applied, and work through it; this becomes a talking point and portfolio asset." That artifact does triple duty: it survives the resume screen, it seeds the technical interview, and it signals the hybrid fluency (domain knowledge plus AI tooling) that Nexford identifies as the "sweet spot where human strengths meet machine strengths."

Reverse recruiting services exist (Shinkarovsky's agency submits 50 to 100 human-crafted applications weekly per client and messages five to ten contacts at each target company) but they carry cost and risk. Zapier's Bonnie Dilber warns candidates to vet providers through independent referrals, Reddit threads, and Trustpilot reviews, and to stay hands-on: "if they misrepresent your experience, then you might just end up spending a lot of money on these services without really getting the results you want." Shinkarovsky himself says anyone already converting at a 10 percent interview rate (one interview per ten applications) in the first month or two likely doesn't need the service. The higher-ROI move is investing that budget in a focused upskill sprint: a 90-day plan with dated milestones: complete a domain-relevant AI course, ship the mini-project, refresh LinkedIn headline and summary to flag AI-adjacent readiness, and schedule three informational interviews per week.

Persistence compounds. The Ohio research notes that "AI still requires a human touch… employers know that human verification is still necessary." Candidates who treat each rejection as data (requesting feedback where possible, adjusting keywords, refining the project narrative) move through the funnel faster than those who spray identical applications. In a market where one in four seekers has been looking for six months or longer, the differentiator is not volume but velocity of iteration.

Where the Market Is Moving

The pressure FINNY AI faces is not isolated. Across the sector, firms are rewriting the rules of recruitment: some by throwing capital at the problem, others by buying entire teams, and a few by reshaping where talent wants to live.

Big Tech has turned poaching into a line item. Meta chief executive Mark Zuckerberg offered $100 million signing bonuses to lure top OpenAI researchers, and reportedly put $250 million on the table for Matt Deitke, a 24-year-old who left a University of Washington computer-science doctorate. Google countered by acquiring Windsurf, the AI coding startup co-founded by Varun Mohan, in a $2.4 billion deal that folded Mohan and his team into Google DeepMind. Microsoft AI, meanwhile, quietly hired two dozen DeepMind engineers. The message is plain: when a single frontier model costs $79 million (GPT-4, 2023) to $192 million (Gemini 1.0 Ultra) to train, a $10 million engineer looks like a rounding error. "If I'm going to spend a billion dollars to build a model, $10 million for an engineer is a relatively low investment," said Alexandru Voica, summarizing the logic that now drives nine-figure packages.

The acqui-hire has become a strategic alternative to individual recruiting. Anthropic absorbed Humanloop (a University College London spinout that had raised $7.91 million from Y Combinator and Index Ventures) bringing aboard its three co-founders, CEO Raza Habib, CTO Peter Hayes, and CPO Jordan Burgess, plus roughly a dozen engineers and researchers. Humanloop had built a reputation helping enterprise customers such as Duolingo, Gusto, and Vanta evaluate, monitor, and fine-tune LLM applications. Anthropic's API product lead Brad Abrams called their "proven experience in AI tooling and evaluation" invaluable for advancing safety and enterprise readiness. The move also came days after Anthropic struck a $1-per-agency first-year deal with the U.S. government's central purchasing arm, a direct undercut of OpenAI's federal pricing. Talent acquisition and go-to-market strategy are now the same motion.

Salary data confirms the inflation. Indeed pegged the average U.S. machine-learning engineer at $175,000 in 2025. In London, principal ML engineers command £140,000 to £300,000. First-party board data from Zero G Talent shows Anthropic advertising roles at $350,000–$850,000 (median $405,000 across 528 salaried roles) and adding 48 openings in the past week alone. Databricks, another heavyweight, listed 47 new roles in the same window with a band of $140,000–$318,000 (median $250,000). These are not outliers; they are the new floor.

Geography is shifting beneath the bidding war. The 2025 Global Talent Competitiveness Index, produced by INSEAD and the Portulans Institute, ranked Singapore first — leapfrogging Switzerland — while the United States fell to ninth, its weakest showing since 2013. Seven of the top ten are high-income European economies; Denmark, Finland, and Sweden all climbed. Singapore's rise was driven by a seven-place jump in talent retention (38th to 31st) and a first-place ranking in formal education and "Generalist Adaptive Skills." Paul Evans, co-editor of the report, argued that economies cultivating "adaptable, cross-functional and AI-literate workforces" will convert disruption into sustained competitiveness. The U.S. slide is a warning: Stanford's Human-Centered AI Institute found that nearly all DeepSeek researchers were educated or trained in China, and more than half never left. Of the quarter who gained U.S. experience, most returned to build in China. Export controls and compute subsidies alone cannot reverse that flow.

Traditional industries are being priced out. Mark Miller, founder of Insurevision.ai, described a "massive opportunity gap" in insurance, healthcare, and logistics: sectors that need AI innovation but cannot match Big Tech compensation. "You can't have one industry hoarding all the talent while others wither," he said. Voica predicts a bifurcation: some engineers will chase the highest salary and accept Big Tech bureaucracy; others will choose startups where equity and ownership compensate for lower cash.

Talent development is also becoming a competitive arena. The inaugural CyberBay conference in Tampa Bay (blending academic, government, military, and industry participants) awarded a $20,000 grand prize to a multi-university team and plans a second edition for spring 2026. Students described the format as forcing them out of comfort zones, sharpening communication, and building networks with professionals and professors. These events are less about hiring today than about shaping the pipeline for 2027.

The sector's response to hiring pressure is no longer just higher offers. It is acquisitions that import entire toolchains, government deals that lock in distribution, geographic arbitrage that follows talent to where it wants to stay, and pipeline investments that try to grow the supply FINNY AI and its peers are fighting over.

The Screen That Learns

The screening pressure exemplified by AI-native firms — multi-stage gates, live technical demonstrations, explicit tests for AI-assisted work — will become the sector baseline, not the exception. HBS researchers document a many-fold increase in applications per role as candidates use generative tools to mass-produce tailored resumes and cover letters that pass semantic matching filters. "How do you filter 1,000 applications when you are expecting 50?" one researcher asked. The answer emerging across firms is more friction: video submissions, timed coding exercises, essay prompts, and sequential interviews that raise the marginal cost of applying. Each layer filters volume but also raises the false-negative rate for genuine talent.

Fraud accelerates the arms race. HBS sources confirm companies have hired candidates who were not the people interviewed: AI-generated video avatars, synthetic voices, and fabricated LinkedIn profiles backed by real Social Security numbers. These actors work remotely, never meeting colleagues in person. Verification will shift toward in-person or proctored checkpoints: live whiteboarding, on-site technical days, and identity confirmation that cannot be spoofed by current generative models. The cost of a bad hire (especially in AI roles where model-access privileges create security exposure) justifies the expense.

The criteria themselves are mutating. Yale's survey of 100 university presidents found critical thinking and complex problem-solving now rank as the most sought capabilities by a wide margin, followed by adaptability, creativity, and technical analysis. Employers "are no longer just looking for workers who can execute tasks. They are looking for those who can exercise reasoning in AI-enabled environments." Screening will weight judgment over syntax: can the candidate decompose an ambiguous problem, decide what to delegate to an agent, and audit the output?

Agentic AI compresses the half-life of any static skill screen. As HBS researchers put it: "What we call augmentation today might be automation tomorrow." A screen that tests prompt engineering in 2025 tests workflow orchestration in 2026 and exception-handling judgment in 2027. Companies will adopt rolling evaluation frameworks (quarterly skill audits, project-based probation periods, and continuous contribution tracking) rather than one-time gatekeeping. Gartner's CHRO survey urges organizations to "shift from supporting role mastery to driving employee versatility," a mandate that reshapes hiring from a point-in-time filter into an ongoing verification loop.

The pipeline tension remains unresolved. Twenty-two percent of CHROs report at least one business leader has stopped hiring for entry-level roles due to AI automation, per Gartner's 4Q25 survey of 110 HR heads. Yet eliminating those roles forces firms to pay premiums for experienced talent hired externally rather than developed internally. Yale's research warns the greatest risk "will not be a sudden wave of layoffs. It will be a labor market in which fewer entry-level jobs are created, making it harder for workers to gain experience and advance over time." Screening practices that only optimize for immediate productivity (filtering out anyone who needs ramp time) accelerate that hollowing.

The sector's next benchmark will be screens that preserve pipeline while proving capability: paid trial projects with defined scope, mentored contribution periods, and structured apprenticeship gates that convert to full roles. FINNY's two open engineering roles — each requiring demonstrated judgment alongside technical fluency — sit at the leading edge of that model. Firms that treat screening as a one-way filter will fill seats but starve their future senior ranks. Firms that treat it as a two-way development contract will build the workforce agentic AI cannot replace.

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