The Batch: 16 Roles, One Week, One Office
Hedge, a wholesale insurance brokerage building its founding team, posted 16 salaried roles in seven days, according to Zero G Talent. The batch spans five wholesale broker specializations — Entertainment, Medical Malpractice, Commercial Auto, Property — plus a founding underwriter and a founding head of partnerships, Zero G Talent's board data indicates. Every listing carries the "Founding" prefix, marking them as early, equity‑eligible positions, Zero G Talent found. Salary bands cluster at $125,000–$250,000 for brokers and the underwriter; Zero G Talent's data shows the partnerships lead lists at $120,000–$250,000. The board median across all 16 sits at $250,000, as Zero G Talent reported.
All 16 require San Francisco in‑office work except the partnerships role, which is remote. That split signals two motions: concentrating underwriting and distribution muscle at a single desk, while building carrier relationships and capacity partnerships that don't demand daily presence.
The compression — 16 roles, one week — is the headline, Zero G Talent's figures put the batch at 16 roles in one week. It implies Hedge either secured a funding milestone that unlocked headcount budget or hit a product‑readiness threshold where the founders can no longer cover the commercial surface area alone. The "Founding" title on every role reinforces the latter: these aren't backfills; they're the first layer of specialists around the founding team.
Entertainment, medical malpractice, commercial auto, and property are four distinct risk classes, each with its own carrier appetite, regulatory nuance, and distribution rhythm. Staffing all four simultaneously — plus the underwriting and partnerships functions to bind them — signals a multi‑line platform from day one, not a single‑line wedge. The tight top‑end bands reflect the market for experienced wholesale brokers who carry their own carrier relationships and submission flow. The remote partnerships lead, priced similarly, signals that partnership revenue carries equal weight.
For candidates, the seven‑day window matters. Applications submitted in the first 48 hours of a batch this size tend to get human review; after that, volume forces triage. The "Founding" label means equity packages are still being calibrated — early applicants have more leverage to negotiate grant size and vesting terms than those who apply once the cap table settles. Hedge is building its commercial engine now, in public, and the window to join as a founding specialist is measured in weeks, not months.
The Market Context: Hedge Fund AI Arms Race (Separate from Hedge)
While Hedge builds its insurance brokerage team, the broader talent market is running hot on AI. Prop‑trading firms and hedge funds have spent two years converting quantitative strategies into generative‑AI plays. Stanford's AI Index shows AI postings shrank from 2.0% of U.S. listings in 2022 to 1.6% in 2023 even as private investment in generative AI alone topped $25 billion — nearly nine times the 2022 figure. The United States captured $67 billion of that, dwarfing China's $8 billion and the U.K.'s $4 billion. Worker access to AI tools jumped 50% in 2025, and Deloitte projects the share of companies with 40% or more of projects in production will double within six months. The bottleneck isn't capital; it's people who can turn models into revenue.
Firms are paying accordingly. Business Insider's 2024 survey of prop shops and hedge funds laid out base‑salary bands:
| Firm | Role | Base Salary Range |
|---|---|---|
| Point72 | AI Product Lead, Investment Services | $400,000 |
| Bridgewater Associates | ML Research Engineer | $210,000 – $350,000 |
| D.E. Shaw | ML Developer | $250,000 – $350,000 |
| Jane Street | ML Engineer / Researcher / Performance Engineer | $250,000 – $300,000 |
| Two Sigma | AI Research Scientist | $215,000 – $250,000 |
| Two Sigma | Quantitative Researcher (ML) | $165,000 – $325,000 |
| Balyasny | Head of Applied AI (named Nov 2023) | Not disclosed |
| Jump Trading | AI Research Scientist | $200,000 – $300,000 |
| Millennium Management | Full Stack Engineer – Infrastructure AI | $175,000 – $250,000 |
| Point72 | ML Researcher | $175,000 – $250,000 |
| Point72 | ML Researcher – Intern (annualized) | $150,000 – $200,000 |
Levels.fyi tracked the premium widening: entry‑level AI engineers earned 8.6% more than non‑AI peers in 2024, second‑level 11.2% more, senior 10.8% more, and staff 11.1% more. Median AI engineering compensation climbed from $231,000 in August 2022 to $300,600 by March 2024. Axiom Recruit's 2026 data puts the new standard for a senior engineer who can optimize inference costs and run retrieval‑augmented generation at scale at $250,000 base.
The comp mix is shifting, too. Axiom found 42% of senior AI specialists now take more than half their total pay in equity or token grants, and 72% of engineers say they'd choose equity upside over a 10% base‑salary bump. In San Francisco, senior AI engineer base pay averaged $252,000 in early 2026 — a 14% year‑over‑year jump. Yet 64% of senior engineers told the same survey they'd take a 15% pay cut for a superior data stack. Average tenure has collapsed to 22 months; replacing one of those engineers costs up to 213% of their annual salary in lost productivity and recruiting fees.
Hedge's board median of $250,000 sits competitive with the lower end of the prop‑shop range but well below Point72's $400,000 ceiling for an AI product lead. Candidates who clear Hedge's insurance brokerage screen will field counteroffers from firms paying 30–60% more in cash, plus equity structures designed to vest over four years in a market where two‑year tenures are the norm.
The Industry Screen: What Hedge Funds Actually Test (Not Hedge)
The research on actual hedge fund interview processes — drawn from quant funds and multi‑manager platforms — describes a machine built for a different skill set than Hedge's posted insurance roles. What follows is the industry‑standard hedge fund screen; Hedge's brokerage roles would test different competencies.
The Multi‑Stage Funnel
QuantInsti and DalOopa both map the typical process at three to five stages. Street of Walls documented a four‑month gauntlet: four 30‑minute "Superday" behavioral rounds (each demanding a long and short pitch plus light accounting), a two‑hour in‑house three‑statement modeling test, a seven‑day take‑home case study with write‑up and model, a panel pitch to five analysts, a one‑hour Calipers psychological assessment, and final 45‑minute conversations with two portfolio managers. DalOopa confirms three evaluation pillars: technical proficiency (modeling, valuation, portfolio metrics), strategic thinking (inefficiency identification, trend capitalization), and interpersonal clarity under scrutiny.
Role‑Specific Technical Gates
Online assessments diverge sharply by seat. Quant Researchers face probability, statistical modeling, math puzzles, and ML modeling logic. Quant Developers get Python/C++ coding challenges, debugging, and system design basics. Quant Traders are tested on mental math, pattern recognition, trading strategies, and backtesting. Data Scientists see Python, ML fundamentals, and exploratory data analysis. Risk Analysts confront risk concepts, scenario analysis, Excel, and SQL. QuantInsti notes these are role‑specific filters, not a universal bar.
Take‑Homes That Mirror the Job
Take‑homes are "especially common" and job‑shaped: Quant Researchers build factor models; Quant Developers construct mini backtesting engines; Data Scientists clean and analyze noisy datasets; Risk Analysts produce mock risk reports or stress‑test scenarios. Follow‑up interviews with hiring managers, team leads, traders, and PMs then press deeper: brain teasers, system design, strategy logic, and finance acumen checks — "How would you hedge a portfolio?", "What is VaR?", "Explain how you'll check cointegration between two time series" — plus behavioral probes on failure and ambiguity.
The Competencies That Decide
AceTheRound's prep guide distills the screen to five observable behaviors: generating an actionable view (not a book report), supporting it with drivers and sensitivities, pressure‑testing downside with explicit invalidation triggers, communicating concisely (conclusion first, two to three key drivers, then numbers and risks), and demonstrating intellectual honesty, separating facts, hypotheses, and open questions in real time. Transtrend's 2024 interview adds cultural markers: extreme enthusiasm for something, comfort with peers smarter than yourself, low‑ego debate that improves the thesis rather than defending a script, and no need for hard KPIs to clarify the job. One candidate who complained the test "was not at all like a typical test for a hedge fund" was not hired.
Strategy Fit as a Hard Filter
DalOopa and AceTheRound both flag strategy fit as non‑negotiable: a long‑duration narrative pitched to a catalyst‑driven seat fails. Market analysis habits must be repeatable: tracking two to three sectors daily, writing short notes on what changed, what the market implied, and the second‑order impact. Valuation is a decision tool, not an output: pick the right anchor (multiples, sum‑of‑parts, unit economics, DCF where appropriate) and define the variables that truly drive the outcome. The pitch structure is rigid: 20‑second headline, two minutes on drivers, one to two minutes on valuation and risk, then open for questions.
The AI Layer in Modern Screens
Citadel's December 2025 disclosure that stockpickers use an internal chatbot to accelerate research, WorldQuant's AI‑driven data restructuring from images and audio, Point72's CTO Ilya Gaysinskiy ramping tech orgs with AI, Bridgewater's AIA Labs replicating the full investment process, and Balyasny's AI bot targeting senior‑analyst grunt work, all signal that fluency with LLM tooling, prompt engineering, and model evaluation is becoming table stakes at the largest platforms. Roughly 80% of Balyasny staff use its internal AI tools; the firm hired a former CIA AI developer as a data science executive in 2024. Around 39% of hedge fund postings on eFinancialCareers already demand Python; the majority are experimenting with ML models.
The screen rewards decision quality over trivia, variant perception over consensus, and coachability under pushback over polished scripts.
For Hedge's actual posted roles (wholesale brokers and underwriters), the screen would shift toward market access, carrier appetite knowledge, submission quality, and loss‑cost modeling. But the hedge fund screen documented above is the one the talent market expects when they see the name.
Candidate Tactics: Engineering the Application
The first filter isn't human. Applicant tracking systems reject roughly three‑quarters of resumes before a person ever sees them, and the recruiters who do review the survivors spend about six seconds scanning each one. Candidates targeting hedge fund AI roles have internalized this math: a generic resume sent to 150-plus postings, the volume forum users report across insurance, banking, and hedge funds since May 2024, yields almost no callbacks. The response has been a shift toward per‑application tailoring, driven by tools that extract exact keywords from each job description and weave them into the candidate's actual experience.
Keyword density has become a calibrated science. Research on quant‑specific ATS systems shows that CVs containing fewer than 12 relevant terms embedded in work experience and project descriptions, not merely listed in a skills block, suffer low pass rates. A weak skills section reads "Python, machine learning, data analysis, financial modelling." A strong one reads: "Python (pandas, NumPy, scikit‑learn, PyTorch, statsmodels, Zipline/Backtrader for backtesting), R, SQL, C++ (basic), Git, LaTeX. Machine learning: gradient boosting (XGBoost, LightGBM), LSTM/transformer architectures, random forests, logistic regression, regularised linear models, NLP (spaCy, HuggingFace transformers, BERT fine‑tuning). Statistical methods: time series analysis (ARIMA, VAR, Kalman filter), Bayesian inference (PyMC), cointegration testing, bootstrap methods." The difference is specificity that survives both algorithmic scoring and the six‑second human scan.
Project portfolios have replaced vague accomplishment bullets. Quant interviewers at Two Sigma and Citadel routinely read GitHub repositories before interviews. A credible profile now shows at least two to three repositories with substantive finance‑related content, clean documented code, README files explaining the hypothesis, data sources, validation methodology, and honest results, including Sharpe ratios, hit rates, and identified failure modes. One documented project format: "Equity Momentum Factor Model | Python, pandas, scikit‑learn | 2025: Built long‑short equity strategy based on 12‑1 momentum signal across US large‑cap universe (Russell 1000, 2000‑2024 daily data). Implemented mean‑variance optimisation with L2 regularisation; achieved Sharpe 0.71 out‑of‑sample (2020‑2024) after 25bps round‑trip transaction cost assumptions. Identified significant momentum decay post‑2021 earnings surprise periods; added earnings announcement buffer leading to 18% reduction in drawdown." The key elements are specific data source, time period, validation approach, honest numerical outcome, and an insight beyond the raw result.
A researcher who presents an ML model with 92% in‑sample accuracy and no out‑of‑sample validation will be dismissed immediately. A researcher who presents a model with 54% directional accuracy but rigorous out‑of‑sample walk‑forward validation and documented Sharpe above 0.8 after transaction costs will generate genuine interest.
Candidates without existing projects are building them in the three to six months before applying. QuantConnect's cloud‑based backtesting platform, Zipline's open‑source library, and Kaggle competitions provide the infrastructure. The López de Prado textbook Advances in Financial Machine Learning serves as the standard reference. Fine‑tuning or prompting LLMs on financial text, such as earnings transcripts, regulatory filings, and news, and evaluating predictive validity has become one of the highest‑signal additions to a quant CV in 2026, even at academic‑project scale.
AI‑assisted tailoring tools have moved from experiment to standard practice. Services like Resume Hedgehog ingest a base resume and a job description, then generate an ATS‑optimized version in under two minutes by mapping the posting's exact keywords into the candidate's real experience. The workflow: paste resume once, paste each job description, answer targeted follow‑up questions that surface the most relevant achievements for that specific role, download a formatted PDF. Users report applying to ten jobs in the time it previously took to tailor one. The same principle drives cover letter generation: feed the job description, resume highlights, and two quantified wins; get a 180‑220 word draft built on a hook‑proof‑close structure; then edit ruthlessly for specificity and voice.
Interview preparation has adopted the same intensity. Candidates run simulated circuits: 15 minutes behavioral, 15 minutes role‑specific scenarios, 10 minutes compensation logistics. Tools like Yoodli and Google's Interview Warmup record responses, flag filler words, and score against the STAR framework. For technical roles, candidates paste recent incidents or datasets and request whiteboard prompts. The goal is a 60‑second "value reel" that ties verified wins to the role's outcomes, rehearsed until it sounds natural.
The pattern is clear: volume alone fails. The candidates advancing through hedge fund screens are those who treat every application as a discrete engineering problem: match the specification, prove the build, document the test results.
Internal Sequencing: Commercial First, Engineering Later
Hedge's 16‑role announcement, all founding‑level commercial positions, signals a company building its go‑to‑market engine before its engineering organization. The roles cluster in wholesale brokerage across four lines, plus a founding head of partnerships and a founding underwriter. None carry engineering, research, or product titles.
That composition matters. In the multistrategy hedge fund world, where Citadel, Millennium, and Balyasny have driven a 90% headcount increase since 2019 while adding $200 billion in assets, hiring typically follows a sequence: investment talent first, then business development to recruit more investment talent, then infrastructure. Citadel's Ken Griffin attributed the firm's record 2023 profitability partly to an "unparalleled" ability to "recruit experienced professionals" and attract "gifted graduates." But those hires were portfolio managers and analysts. Hedge's current batch looks different.
The founding wholesale broker roles suggest Hedge is staffing distribution channels for specific insurance lines. A founding head of partnerships (remote, $120,000–$250,000) implies platform or channel strategy. The founding underwriter role points to risk selection capacity. Together, they form a commercial spine: underwriting authority, distribution reach, and partnership leverage. Engineering hires, if they follow, would likely serve this commercial core, building pricing tools, submission pipelines, or capacity management, rather than leading a product roadmap independently.
Market data supports this sequencing. Qube Research & Technologies, another quantitative firm scaling rapidly, made a "high profile hire from Jump Trading" for its hardware division only after its broader hiring spree added staff in the hundreds. Balyasny added six business development professionals in 2024, including three managing directors, to "facilitate hiring" of portfolio managers. The pattern: commercial and recruiting infrastructure precedes technical expansion.
For Hedge, the 16 roles represent a commercial foundation. The board data shows no engineering requisitions in the same window. If the company follows the multistrategy playbook, product and engineering hiring will accelerate once distribution capacity is proven, when the brokers and partnerships generate submission flow that demands automation. Until then, roadmap acceleration remains a second‑order effect, gated on commercial traction the new hires are hired to create.
The tension is real: the research shows a commercial hiring burst, not an AI‑engineering one. If Hedge's screen rewards specialized AI talent, that signal isn't visible in the 16 roles posted. The screen may exist for roles not yet advertised, or the "AI‑focused" label may describe the product these commercial hires will eventually sell. Either way, the internal impact right now is distribution build‑out, not engineering scale‑up.
Out of Scope: Why the Name Misleads
The name "Hedge" carries heavy baggage. In finance, a hedge is an investment position intended to offset potential losses or gains incurred by a companion investment, a risk‑reduction strategy built on derivatives, forwards, and options contracts that has spawned a global market in products to hedge financial market risk over the last fifty years. In agriculture, a hedgerow is a line of shrubs and sometimes trees, planted and trained to form a barrier or mark a boundary; England alone was found to have 390,000 km of hedgerows as of 2024, per a UK Centre for Ecology & Hydrology lidar study. The word traces to Old English hecg, meaning any fence, living or artificial, with the verb sense of "insure oneself against loss" appearing in the 1670s. None of this history, terminology, or technique maps to what the company Hedge is currently screening for.
The first‑party board data shows Hedge's 16 newest roles carry titles like Founding Wholesale Broker (Entertainment, Medical Malpractice, Commercial Auto, Property), Founding Head of Partnerships, and Founding Underwriter, with salary bands of $125,000–$250,000 and a board median of $250,000. These are insurance brokerage and underwriting positions, not machine‑learning engineering or research roles. Yet the hiring surge described across this article centers on an AI‑focused screen. The tension is real: the live postings read like a specialty insurance build‑out, while the narrative framing treats Hedge as an AI talent magnet. Candidates should treat the posted roles as the ground truth, the screen will evaluate for the competencies those roles demand.
What those competencies are not: familiarity with delta hedging, variance swaps, or collar strategies. No interviewer will ask you to price a put option on crude oil futures, the instrument Southwest Airlines used to save heavily after the 2003 Iraq war and Hurricane Katrina, nor to explain why a farmer locking in a forward price gives up the upside of a price increase. The mechanics of basis risk, counterparty risk, and the cost of carry (hedging isn't free; monthly payments add up, and if the flood never comes, the policyholder gets nothing) are irrelevant. So is the botanical side: hedgerow dating, soil stabilization, nesting‑bird cutting windows (March–August in Britain), or the £70 million UK farmers received for tending hedgerows in 2024. The company's name is a brand, not a curriculum.
What the screen likely does reward, inferred from the role titles, is wholesale brokerage fluency: market access, carrier relationships, submission management, and the ability to structure complex risks in entertainment, med‑mal, commercial auto, and property lines. The "Founding" prefix signals early‑stage, zero‑to‑one building: pipeline creation, partnership negotiation, underwriting authority design. If an AI layer sits atop this, automated submission triage, risk‑appetite matching, pricing augmentation, the screen will test for product‑minded engineering: can you ship reliable software into a regulated, relationship‑driven workflow? Can you translate broker feedback into model features? Can you operate within compliance constraints that don't exist in a research lab?
Candidates who lead with CFA terminology or agricultural history will signal a category error. The winning applications will mirror the job specs: brokerage outcomes, partnership deals closed, underwriting guidelines authored, systems built for capacity‑constrained markets. Hedge's screen filters for that execution profile. Everything else, financial theory, botany, etymology, is noise.
Working in frontier tech? Zero G Talent tracks the openings: see every open Hedge role, browse frontier tech jobs, the companies hiring, and the people building the field.