How Work Gets Done
Autonomous Technologies Group, an eight-person startup founded in 2025 by the team that sold Paperspace to DigitalOcean for $111 million, is building what it calls a financial superintelligence, an autonomous agent that must clear SOC 2, FINRA, FDIC, and SEC compliance before its early 2026 launch.
ATG's culture is shaped by the constraints and ambitions of its small size and the breadth of its technological scope. Public access is slated for early 2026, starting with iOS. By launch, the platform must be SOC 2 compliant, FINRA-licensed, FDIC-approved, and SEC-compliant. That regulatory surface area (banking, securities, data privacy) sits atop a stack the company's job board lists across nine open technical roles with salary bands clustered between $115,000 and $300,000.
Decision-making runs through the founders, Dillon Erb and Daniel Kobran. LinkedIn shows 2–10 employees; a January 2026 report put headcount at 15, including engineers from Meta, Spotify, and American Express. With roughly eight people, every engineer touches multiple layers. The board's role listings (Quantitative Systems, Data Platform, Backend, Research Engineering) read less like org-chart lanes than like a checklist of capabilities the whole team must cover.
FINRA licensing and SEC compliance impose hard external deadlines. The Deloitte framework for AI-first companies describes an "autonomous operating model" running in "shorter, AI-informed cycles" with "leadership as system orchestrator." At ATG's scale, there is no separate platform team to absorb infrastructure pain. Coordination across domains happens in the codebase and in the compliance artifacts. The "human/agent collaboration" that Deloitte flags as a pillar of AI-first organizations is, at this stage, the team collaborating with its own future automation: every internal tool is a candidate for the agent that will eventually serve customers.
Values That Function as Filters
Autonomous Technologies Group operates from a mission that doubles as a constraint: "High quality and personalized financial advising shouldn't be a luxury reserved for the ultrawealthy," founder and CEO Dillon Erb said in the company's January 2026 launch announcement. The explicit target — helping "millions of people across the income spectrum 2x their retirement savings" — frames every product and hiring decision as a democratization problem, not a premium-service play. That framing traces to Erb's personal catalyst: after selling Paperspace, he tried to organize his own finances and found both DIY spreadsheets and traditional advisors broken. "Run that math over multiple decades and you're handing over half your net worth to fees," the company's Y Combinator page notes.
The company codifies its approach in three core objectives published on its site: deeply understand the market, deeply understand an individual's financial life, and advance the system's representational capabilities. "Everything the system does follows from these three commitments," the documentation states.
That research bias shows up in regulatory posture. ATG registered as an investment advisor (RIA) under the SEC and FINRA before launch, and it publishes an "Investment Doctrine": the guidelines governing how the system makes suitable recommendations across the full range of financial decisions.
The technical philosophy extends to what the company refuses to build. The Y Combinator page is blunt: "Robos were supposed to fix this, but they were designed in a pre-AI world. In practice they just manage cookie-cutter portfolios." The critique lists specific gaps: no coordination across 401(k), taxable accounts, mortgage, equity, and cash flow; no help with real decisions like job offers, windfalls, or concentrated stock; no delivery of institutional strategies. The result: "even after a decade plus, the biggest robos are still under $100 billion AUM while old-school advisors like Raymond James and Edward Jones sit on trillions."
ATG's operating principle is to solve the coordination problem robos ignored: a single system that sees the whole balance sheet and acts on it. The eight-person team includes research scientists, quantitative systems engineers, and backend engineers, roles mapping to the three core objectives rather than traditional product/engineering/design splits.
The long-term vision makes the operating principle explicit: "Today, Autonomous is an application that onboards you into a system that learns your financial life completely and guides you with precision. Over time, it will act on your behalf — a financial superintelligence working at all times, whether or not you open the app." That phrase is the tell. The company is building an autonomous agent, not a dashboard.
What the Hiring Bar Selects For
| Role | Salary Band |
|---|---|
| Backend, Research Engineering, Quantitative Systems | $180k–$260k |
| Data Platform | $150k–$275k |
| Research Scientist, Client Director | $200k–$300k |
| Aggregate (9 roles) | $143k–$300k, median $260k |
The job postings on Zero G Talent's board tell a clearer story than any mission statement. ATG lists nine open roles, all carrying the "Member of Technical Staff" title except for a Client Director and a Research Scientist. The compensation level, at a company founded in 2025 with roughly eight people, signals an explicit preference for engineers who can operate without scaffolding.
The title "Member of Technical Staff" is itself a filter. Borrowed from research labs and elite engineering organizations, it denotes an individual contributor expected to own problems end-to-end: architecture, implementation, testing, and often the product conversation that precedes them. ATG's variants (the same four domains listed earlier) reveal the technical surface area. A candidate who has only ever worked on one layer of that stack will struggle to demonstrate the breadth the role demands.
The Research Scientist posting at the top of the band suggests the company treats novel algorithm development as a first-class engineering discipline. The Client Director role, sitting at the same band, indicates commercial deployment is already a parallel track. Geography reinforces the signal: roles are anchored in New York City and San Francisco with U.S. remote eligibility.
The board data shows no junior titles, no "associate" or "intern" designations, and no roles focused solely on tooling or infrastructure maintenance. Every opening is for a domain that directly advances the autonomous system.
What the postings omit is equally telling. There is no mention of specific frameworks (ROS 2, PyTorch, JAX), simulation platforms (CARLA, Isaac Sim, custom), or sensor suites (lidar, radar, vision). The absence suggests the interview loop tests fundamentals (estimation theory, real-time systems, distributed systems correctness, data pipeline reliability) rather than framework fluency.
The eight-person team size creates a secondary filter: every hire must increase the team's collective bus factor, not decrease it. That means demonstrated ability to document, test, and hand off work, habits optional in larger organizations but existential here.
The Silence Around Employee Reviews
The company was founded in 2025 and employs roughly eight people. Platforms that aggregate workplace sentiment typically require a critical mass of reviewers before they publish anything, and an eight-person team operating for less than a year simply has not generated that footprint. Anyone claiming to quote "employee reviews" for this company is fabricating.
The broader market context explains why. The tech layoff wave that accelerated through 2024 continued into 2025: more than 150,000 cuts across 549 companies last year, per Layoffs.fyi, and over 22,000 more in the first months of this year alone. February 2025 saw 16,084 reductions in a single month. Companies like Intel (21,000+ cuts), Microsoft (9,000), and Cruise (half its workforce) have been resetting headcount aggressively.
MIT Sloan data on agentic AI adoption (35% of organizations already deploying, another 44% planning to within two years) puts ATG in a category the market is still figuring out how to value. The 76% of executives who now view agentic AI as more like a coworker than a tool suggests the company's technical bet is directionally aligned with where enterprise buyers are moving.
For a candidate evaluating this company, the lack of reviews means you cannot outsource due diligence. The salary bands say they can afford good people. The role definitions say they need generalists who go deep. The rest is unverified — and at this stage, that is the only honest answer.
Who Thrives and Who Burns Out
ATG's scope includes a live model of a user's entire financial life (assets, liabilities, cash flow, taxes, including off-platform holdings like rental properties and startup equity), a conversational interface for questions and scenario planning, and hyper-personalized direct indexing optimized at the individual security level; compressed into a team this small, this scope means every hire operates at the intersection of multiple disciplines.
Candidates who thrive share three traits. First, genuine breadth: the board shows roles that in a larger organization would sit in separate departments (quantitative research, data infrastructure, backend engineering, client strategy) all recruited at the same $180k–$300k band (median $260k). Second, comfort with undefined process: at eight people, there is no "platform team" to file a ticket against. Third, tolerance for the financial domain's regulatory and precision constraints. The LinkedIn description emphasizes "how money gets managed...", a remit leaving little room for the "move fast and break things" posture that works in consumer social. A mistake in direct indexing execution is not a rolled-back deploy; it is a tax event or a compliance violation.
The burn-out profile is the mirror image. Specialists who need a clear swim lane will find the lack of handoff points exhausting. Candidates who equate early-stage with "pre-product-market-fit chaos" misread the domain: financial autonomy requires audit-grade correctness from day one. People who rely on external structure (sprint ceremonies, dedicated QA, a product manager who writes specs) will spend more energy building that structure than shipping. The salary band is competitive with Series A/B companies, but the equity upside and role breadth are the actual compensation.
Geography adds a subtle filter. Roles are listed in San Francisco, New York City, and Remote (US). The remote option exists, but the team's density in two expensive hubs suggests in-person collaboration is the default mode for the hardest problems.
The company's own line — "We think this is how money gets managed going forward" — is both the mission and the filter. When the agent goes live in early 2026, the first user it onboards may be the team itself.
Working in frontier tech? Zero G Talent tracks the openings: see every open Autonomous Technologies Group role, browse frontier tech jobs, the companies hiring, and the people building the field.