The Rhythm of a Fifty-Person Team
Veryfi has operated since its 2017 founding with a 54-person team spanning San Mateo, Medellín, Sydney, and a handful of home offices. The company doesn't run on process documents. It runs on the rhythm its founders set and the coordination habits that survive time zones.
As of April 2025, headcount sat at 54, per Highperformr data. The careers page promises a "fast-paced, fun environment where you can make a real impact." Employee testimonials there name "quick iteration," "continuous feedback," and a "green light culture" that encourages initiative: engineers start and deploy projects to production without layers of approval. Founders work directly with their teams. Ernest Semerda (CEO) and Dmitry Birulia (CTO) remain the technical and product anchors; Mark Gisi leads revenue and Jason Lee leads finance, a leadership slate with no turnover in the past year.
Coordination is hybrid by default. Eight of 13 open roles listed in May 2026 were tagged remote; the rest clustered in San Mateo, New York, San Francisco, and Medellín. The San Mateo office at 210 South B Street still hosts daily shared lunches (a ritual the culture page calls a collaboration mechanism) along with weekly hikes and seasonal baseball outings. The Medellín and Sydney offices, plus the distributed remote contingent, rely on asynchronous handoffs and overlapping hours the company does not publicly document. Employees describe a pace where "you need a team of people who are capable of adopting change and making the best out of it," a reflection of the coordination load that falls on individuals.
Decision-making sits close to the code. The "green light" framing means engineers and product people propose, scope, and ship features without a formal gate committee. One testimonial notes deploying multiple projects to production during their tenure, each used by thousands of users daily. That pace is enabled by a stack built since 2017: proprietary LLMs with vision models, an OCR API, and the Veryfi Lens mobile SDK, infrastructure that lets small teams push updates to document extraction, fraud detection, and on-device field validation without waiting on external vendors. The launch of On-Device Field Detection, which validates vendor, date, and total on the receipt confirmation screen before any server call, shipped from this same small group.
The trade-off shows in hiring. Recent postings skew heavily toward senior IC roles: Senior Python Engineer, Sr. Machine Learning Engineer, Data Scientist, Senior ML Engineer, Sales Engineer, all posted the same day in May 2026. The most-hired roles over the prior 90 days were two Sales Directors, a Technical Account Manager, a Technical Customer Support T2, a Content Marketing Intern, and a single Machine Learning Engineer. The ratio suggests a team that promotes from within or expects new hires to operate at senior autonomy immediately, with minimal onboarding scaffolding.
What holds it together is not a framework but a density of context. The founders still review product direction; the head of revenue and head of finance sit in the same loop; engineering leads own their surfaces end to end. In a company this size, the "operating rhythm" is essentially the founders' calendar made visible. The current leadership has held steady, but the structure remains personality-dependent.
Machine Automation as a Founding Principle
Veryfi's operating DNA bears a distinctive stamp for a Silicon Valley AI company: an insistence on 100 percent machine automation. The founders state this plainly — "Our 100 percent machine automation is what makes us unique" — and the principle ripples through every priority the company sets. Veryfi's OCR API and Lens SDK extract data from receipts, invoices, and time-and-materials documents with no human reviewers in the loop. The company calls the alternative a "somewhat poorly kept secret" in pre-accounting software: many competitors marketed as AI still route documents to human extraction teams. Veryfi treats that model as a trust violation — humans make mistakes, get tired, and introduce privacy risk. Machines don't sleep. The principle is non-negotiable: if the model can't hit the accuracy bar unaided, the product doesn't ship. That standard has shaped engineering culture toward precision over breadth. The team builds native iOS and Android apps (not wrappers) because field crews in construction, real estate, and healthcare work on phones, not desktops. Seven in ten construction workers are mobile; the product meets them there.
Customer-centricity traces back to Y Combinator, where the team was advised to speak with users daily. That habit stuck. Net Promoter Score is measured regularly. The partnership with Intuit (QuickBooks' parent) is framed explicitly as values alignment: "shared values about the Firm of the Future… well grounded in first principles and an aligned culture." First-principles thinking appears repeatedly in founder statements: bookkeeping is procrastinated because it drains willpower; automate the drain, unshackle the mind. The product roadmap follows that logic: voice entry, geofencing, floor-by-floor detection, each feature removing a manual step for a field worker who has no time for a desktop.
Trust is the currency the company hoards. Encryption standards (TLS 1.2, AES at rest and in transit), MFA, biometric one-way encryption, and compliance with GDPR, HIPAA, and CCPA are table stakes. The deeper commitment is architectural: no human augmentation anywhere in the workflow. That choice eliminates a whole class of data-access risk. It also means the engineering bar for model quality is the product bar — there is no fallback team to clean up errors.
The Interview Bar: Proven Judgment Over Raw Potential
Veryfi's interview process moves fast. Glassdoor shows five to seven interview write-ups across regions, though too few to form a reliable signal on their own. The company's stated values (Empathy, Trust, Trailblazer, Have Fun) appear on the careers page next to language about "liberating human potential by making data more accessible and actionable" and "nurturing an atmosphere of joy and enthusiasm fuels creativity."
What's clearer is the geographic spread: roles sit in San Mateo, Medellín, and Sydney, with remote eligibility noted on several current postings. The hiring bar selects for people who can deliver senior-level output across time zones without daily sync rituals: autonomous, configuration-fluent, business-literate, and comfortable staying hands-on.
| Role | Salary Band |
|---|---|
| Data Scientist - Fraud Detection | $160k–$210k |
| Account Manager - Growth Accounts | $150k–$183k |
| Senior ML Engineer | $100k–$160k |
| Full-Stack Software Engineer | $120k–$160k |
The Evidence Base: What It Doesn't Show
The public record on Veryfi's employee experience is thin and heavily filtered. Glassdoor lists 10 anonymous reviews, a sample too small for statistical claims about sentiment but large enough to confirm that people have written about working there. The reviews themselves sit behind Glassdoor's login wall; their content, ratings, and dates are not reproduced in the research available to this article. What we can say: the platform exists as a venue where current and former employees have chosen to post, and the count (10) suggests a very small team or low review participation, consistent with reported headcount.
The company's own careers page publishes six first-party statements framed as employee voices, attributed by name and role: Hoanh (Software Engineer), Matt (Revenue Operations Manager), Fenton Bear (Head of Human Resources), Helen (Senior Customer Success Manager), Zain (Business Development, EMEA), and Mauricio (Sales Director LATAM). They describe "quick iteration and continuous feedback," a "green light culture that empowers team members to leverage their experience to solve meaningful problems and build for the future," and fulfillment when "customers recognize our contributions to their success." They read as curated testimonials rather than spontaneous reviews, and they align with the autonomy language the company uses to describe its operating style.
No critical or negative employee accounts appear in the research corpus: no Glassdoor excerpts citing burnout, management issues, compensation complaints, or coordination failures. That absence could mean the reviews are broadly positive, or it could mean the sample is too small and self-selected to surface dissent. With only 10 reviews on a platform that tends to attract polarized responders, either reading is speculative.
Zero G Talent's first-party board data shows 15 salaried roles posted with bands ranging from $100k to $210k (median $130k), including positions like Senior ML Engineer, Data Scientist - Fraud Detection, and Full-Stack Software Engineer. The existence of multiple remote-eligible listings at senior levels suggests the company is hiring into the autonomous, high-ownership model described in its own materials. But the board data captures employer intent, not employee experience.
The gap between what the company publishes about itself and what independent reviewers have actually written — and whether those 10 Glassdoor reviews corroborate or contradict the "green light" narrative remains an open question. Anyone evaluating Veryfi should read the Glassdoor reviews directly, note their dates, and weigh the anonymous accounts against the named testimonials on the careers page. The evidence base is simply too narrow to declare a consensus.
The Profile That Stays and the One That Leaves
The profile of a Veryfi employee who stays and grows reads like a mirror of the company's operating DNA: autonomous, high-ownership, comfortable in a small, fast-moving team, and fluent in the rhythms of a predominantly remote and hybrid workforce. The board data (15 salaried roles across data science, machine learning, full-stack engineering, and account management, median band $130k, ranging $30k–$174k) shows every listing expects the hire to operate with minimal scaffolding. A Senior ML Engineer or Data Scientist - Fraud Detection posting doesn't come with a detailed spec and a product manager checking in daily; it comes with a problem space and the expectation that you'll define the approach, ship it, and iterate.
People who thrive here share a cluster of traits. They communicate well in writing, not because it's a stated value but because a remote-first, hybrid team across time zones makes async clarity the only way work moves. They self-prioritize ruthlessly; in a small team there's no product ops layer to triage your queue. They treat ambiguity as raw material, not a blocker. And they accept coordination overhead as the tax on flexibility: the same remote policy that lets you work from anywhere also means you're the one who has to schedule the sync, write the update, chase the decision.
Who burns out? The pattern inverts cleanly. Engineers and scientists who need a ticket groomed before they start, who wait for design reviews to unblock them, or who measure productivity by meetings attended will stall. Managers accustomed to a layer of program managers, technical writers, or dedicated QA will find the gaps glaring. People who recharge on daily in-person energy — the hallway conversations, the whiteboard sessions, the lunch-table osmosis — tend to drain fast in a culture where those moments are scheduled, not spontaneous. And anyone who conflates "remote-friendly" with "low accountability" discovers quickly that the opposite is true: ownership is the only visibility mechanism, and invisible work doesn't exist.
The hiring bar selects for the first group implicitly. The roles on the board (Sr. Machine Learning Engineer, Full-Stack Software Engineer, Account Manager - Growth Accounts) all sit in functions where output is measurable and ownership is binary. You shipped the model or you didn't. The account grew or it didn't. That clarity attracts people who want their work to speak and repels people who want their presence felt. The coordination challenges the main theme flags aren't bugs to be fixed; they're the filter. If you can't work through them, you don't last. If you can, the flexibility is real, the scope is yours to define, and the team is small enough that your fingerprints are on the On-Device Field Detection launch — and on whatever ships next.
Working in AI? Zero G Talent tracks the openings: see every open Veryfi, Inc. role, browse AI jobs, the companies hiring, and the people building the field.