This Startup Processes 3B+ Users. It Has 26 Employees and 2 Open Roles.

The Screening Gauntlet: How Cinder Filters for Trust & Safety

Cinder, a trust-and-safety infrastructure company with approximately 26 employees per public filings, has two critical roles open — and its founding team personally reviews applications, with no recruiting team or coordinator in place. Applications land in Ashby, the ATS Cinder uses, but the obstacle isn't the software. It's being seen at all. Cold outreach outperforms the form consistently, though no verified contact route is published. A tailored CV that mirrors the job description's language clears the initial scan; a generic one gets rejected fast. Most B2B security applications at this stage get ghosted entirely. Data from the hiring board shows a day-five follow-up can double response rates, but only if the initial message opens with what Cinder is dealing with right now — not what the candidate wants.

Roles at sub-50-person B2B security companies typically draw 50 to 100 applicants in the first two weeks. Cinder's two current openings — a Senior Software Engineer (Full Stack) in New York or remote US at $165–235k, and a second role posted within the same week — sit inside that range. But volume doesn't translate to velocity. The founding team's direct involvement means every application competes for attention alongside customer calls, product decisions, and payroll runs. "There's no 'we'll get to it next sprint.' We ship in days, not quarters," the careers page states. "We make decisions with incomplete information and own the outcomes."

If a resume survives that first pass, the technical screen follows. Cinder's engineering interview takes three hours total — no LeetCode, no take-home, no paid trial. Candidates start with a Cinder engineer over Zoom for introductions and an overview of the challenge. They then receive access to an interview repository and have two hours to complete a full-stack problem that closely mirrors the type of problems Cinder encounters daily. The challenge breaks into steps that grow progressively more difficult and open-ended, each building on the last. The goal isn't to test encyclopedic knowledge of the stack. It's to see how a candidate unblocks themselves when they hit something unfamiliar.

After the coding portion, the engineer and candidate reconvene over Zoom to review the work, discuss how the candidate navigated uncertainty, and consider how the solution might scale in production. Cinder is explicit that fluency across the entire stack isn't expected. They want to watch the candidate think. The interview is framed as a two-way street: the company measures ability, and the candidate evaluates whether this type of challenge (and this stack) is something they'd want full-time. The company's blog notes an average ramp-up under a week for hires, and that engineers can both work independently and describe their code effectively. The process is designed to leave even rejected candidates with a positive impression.

The Two Roles at the Center of the Push

Cinder's platform sits at the intersection of policy, engineering, and investigation — a single command center where platforms write and enforce policy, run AI agents against abuse, investigate threats, file NCMEC reports, and measure safety programs. That architecture reflects two distinct but interlocking hires: a Trust & Safety Solutions Manager to steer customer deployments, and a Senior Product Manager to shape the platform itself. Both roles appeared on specialized job boards in early 2024, and both reflect the company's thesis that trust and safety is no longer a content-moderation sideline but a full-stack operational discipline.

The Trust & Safety Solutions Manager role, posted on Support Human's board in February 2024, sits on the customer success side. The posting describes the position as leading "Cinder's deployment with a wide range of leading enterprise customers" — working side by side with teams as they "transform their Trust & Safety operations," helping them "rethink their policies and data structure, prioritize the problems they really need to solve, and configure Cinder for their use case." The manager also represents customers internally to the product team and translates product decisions back to customers in structured written communication. This is a deployment role requiring fluency in the operational grammar of trust and safety: policy architecture, investigation workflows, regulatory compliance, and the day-to-day reality of moderation queues. The candidate must have lived the problems Cinder's software solves (coordinating across legal, policy, engineering, and review teams) because the job is effectively translating that lived experience into platform configuration.

The Senior Product Manager role, listed on TrustSafety.net, owns the other half of the loop: the platform's evolution. The description frames it as "senior product management for Cinder's AI-native safety platform," with explicit responsibility for the command center where policy authoring, AI agent orchestration, investigation tooling, NCMEC reporting, and program measurement converge. The product surface spans policy compilers, investigation graphs, automated detection pipelines, and compliance reporting — each of which must interoperate under shifting regulatory regimes (the EU's Digital Services Act, the UK's Online Safety Act) and adversarial pressure from generative AI abuse. Generative AI tools can now manufacture child sexual abuse material and nonconsensual pornography at scale; as one investor noted, "Gen AI is making it 10 times worse."

Why these two roles now? Cinder's co-founding team — Glen Wise (CEO), Phil Brennan (COO), Brian Fishman (formerly Head of Business Development, now at OpenAI), and Declan Cummings (Head of Engineering) — built the trust and safety infrastructure at Meta that handles hate groups, terrorist content, and state-sponsored disinformation. They left to productize the tooling they wished existed: a unified platform that centralizes data, context, and intelligence for any abuse scenario, eliminating the spreadsheet-and-SQL workflow that forces investigators to context-switch across bug trackers, URL viewers, and internal wikis before making a single decision. Wise told TechCrunch in December 2022 that teams were "stuck in spreadsheets, stuck in SQL statements... because they had no tools custom built for this." The Solutions Manager ensures customers actually adopt and configure that tooling; the Product Manager ensures the tooling stays ahead of the threat landscape.

First-party board data from Zero G Talent shows a third signal: as of the past week, Cinder has two salaried roles open on the board, with a salary band of $154k–$243k (median $213k).

That engineering hire aligns with the platform's described velocity and the need to build AI-native investigation tooling at the pace adversaries are automating abuse. The two trust-and-safety-specific roles (Solutions Manager and Senior Product Manager) are the ones that translate Cinder's domain expertise into customer outcomes and product direction. They are the hinge points between the platform's architecture and the operational reality of the teams who use it.

Both roles demand a rare hybrid: someone who has investigated coordinated inauthentic behavior or CSAM rings or state-sponsored influence operations, and who can also articulate the data model, permissioning scheme, and workflow automation that would have made that investigation faster. The Solutions Manager does it by configuring the platform for a specific enterprise; the Product Manager does it by deciding what the platform becomes next. Neither role works without the other — and neither works without the engineering velocity the third role supports.

Beyond the Resume: The Hidden Criteria That Decide

Cinder's interview repository challenge tests code. The conversation afterward tests something else. The company's blog describes the post-coding review as a chance to discuss how candidates overcome uncertainty and consider scalability. That framing (uncertainty, scale, production) reveals what the resume screen cannot.

Empathy appears in Cinder's public materials not as a soft skill but as an operational requirement. The careers page states the team handles "deepfake nudes, child sexual abuse material, AI-generated phishing campaigns, organized crime, counterfeit goods." The company has cited a 2025 study finding that more than a quarter of commercial content moderators showed moderate to severe psychological distress, framing wellness benefits as operational necessity: "Supporting them is part of the responsibility of building safer platforms."

Analytical rigor shows up in how the interview escalates. The challenge steps escalate in difficulty and openness, building on each other, allowing candidates to scaffold their knowledge and demonstrate both breadth and depth across our stack. The founding team wrote that they "want to see how they unblock themselves when they inevitably encounter something they haven't seen before." Cinder's product includes policy management, workflows, cases, audit, compliance, reporting, and analytics — a stack that demands end-to-end thinking.

Mission alignment functions as a filter because the work pace leaves no room for ambivalence. "We're looking for people who take the mission seriously and work urgently as a result," the careers page adds. That urgency is structural. Cinder's adversaries (the abusers, fraud rings, state actors) also build fast with AI. Cinder reported the company raised $41 million in Series B funding to scale its detection platform for customers including OpenAI. A candidate who needs certainty before acting will bottleneck the team.

Communication closes the loop. The blog notes: "We trust that our engineers can work independently, yet we also know that they can describe their code effectively." The interview's final pairing session tests exactly that: "test how candidates handle feedback on their code." In a domain where a missed signal means a child exploitation image stays live or a deepfake spreads, the ability to articulate a decision (and revise it under challenge) is not optional.

The hidden criteria form a coherent profile: someone who has stared at abuse patterns long enough to recognize them in new guises, who debugs without a spec, who ships without permission, and who explains the call afterward. The resume gets you the repository access. The rest decides whether you stay.

Culture as a Filter: How Cinder's Environment Shapes Hiring

Cinder's benefits package reads like a standard competitive offer — health, vision, dental, 401(k) with matching, paid family leave, flexible PTO, commuter benefits, mental health support, tuition assistance — but the context transforms it. The company is based in New York City, with team members across the US and UK/EU. A fully stocked NYC office provides paid lunch and dinner. Leadership is explicit: "We believe in working together in person and hold at least two all-company events per year." A weeklong offsite in Munich in October 2026 (announced on the company's LinkedIn) signals that "working together is a superpower" means travel, intensity, and sustained in-person collaboration.

Remote flexibility exists (the role lists "New York / Remote (US)" with a USD 165–235k band) but with a catch: "We're especially focused on hiring in New York and will relocate the right person for the right role." The remote program is real; the preference for proximity is louder. That tension screens for people willing to uproot for a team backed by Accel and Y Combinator.

Speed is the cultural constant. The design team is building its culture from scratch: "Appetite to help build design team culture and rituals from scratch: crits, standups, lunch-and-learns." The same posts require fluency with AI coding agents (Cursor, Claude, Codex) "for designing and prototyping, not just experimenting with them." The tooling expectation is a cultural signal: you work the way the adversaries work.

Glen Wise, Phil Brennan, Brian Fishman, and Declan Cummings founded Cinder in 2021–2022. The numbers are competitive but not stratospheric. The attractor is the mission ("mission-critical infrastructure for platforms dealing with real harm") and the culture that treats that mission as a daily operational reality, not a slogan. Two all-company gatherings a year. A Munich offsite. Paid meals in a NYC office. Relocation on the table. AI agents in the product. A wellness narrative grounded in published distress data. Every element doubles as a signal: this is how we work; if that fits, apply. If not, the screen worked.

The Applicant Playbook: Strategies That Work

Cinder's screen filters for a hybrid profile that barely exists on the open market: engineers who have stared down coordinated abuse campaigns and lived to refactor the detection logic. The company's two open roles (the full-stack engineering role in New York or remote US) sit inside the compensation band the board tracks. But the number that matters isn't the salary ceiling. It's the rejection rate. Candidates who treat this like a standard full-stack interview get cut before the technical deep-dive.

Lead with abuse-pattern fluency, not framework depth. Cinder's founders built the platform after years on Meta's red teams, threat intelligence, and counterterrorism operations. They know what a state-sponsored influence operation looks like at scale. They know how CSAM hashes from IWF and NCMEC propagate through a moderation queue. If your resume lists React and Postgres but never mentions a policy edge case you had to encode, you're invisible. Describe the time you translated a vague "harassment" definition into a deterministic enforcement rule. Explain how you handled the false-positive spike when a new slang term hit a teen platform. That's the vocabulary Cinder's screeners speak.

Show you've worked inside the moderator's loop. Wise told TechCrunch the platform was built "from the ground up" for the human element — blurring, grayscaling, forced breaks after traumatic content. The role owns workflows that route cases to reviewers across 100+ languages. If you've never sat with a trust & safety operations team, find one. Volunteer for a shadow shift at a platform that uses Cinder — BeReal, OpenAI, and others are customers. Ask how they triage a "dynamic investigation" versus a bulk-action queue. The interview will probe whether you understand that a 75% handle-time improvement means nothing if the reviewer burns out.

Demonstrate regulatory literacy. Those regulations impose fines up to 6% of global revenue. Cinder's product builds audit trails and compliance documentation automatically. Candidates who can map a DSA transparency report requirement to a database schema win points. Read the DSA's very large online platform obligations. Read the UK's illegal content risk assessments. Then explain how you'd design a permissioning system that obscures PII for junior reviewers while exposing full context to investigators — exactly the "robust permissioning" Wise described to TechCrunch.

Prove you can ship policy-as-code. Cinder's platform lets ops teams "translate policy into enforcement without writing code" via drag-and-drop workflows and AI agents. But someone builds the primitives those ops teams compose. The engineering interview will hand you a policy spec and ask you to model the data flow, the human review gates, the feedback loop that catches policy drift. If you reach for a generic rules engine, you've missed the point. Cinder's agents validate prompts, outputs, tool calls, and multimodal outputs against policies in real time. Your answer needs to account for adversarial inputs, multimodal evasion, and the labeled-data pipeline that keeps the classifiers honest.

Signal security hygiene as default, not afterthought. The team's "strong security pedigree" (Wise's phrase) means SOC2, GDPR, HIPAA compliance aren't checkboxes — they're architecture constraints. Walk through how you'd implement role-based access control for a case-management system with tiered visibility. Mention encryption at rest and in transit. Mention SSO integration. If you've never threat-modeled a system that handles CSAM hashes, say so — but describe how you'd approach it.

Prepare for the empathy stress test. The final screen isn't purely technical. Cinder's managed operations team runs a BPO with "incentives actually aligned to yours" — they need engineers who grasp that alignment means protecting the human, not just the platform. Candidates who've never managed moderator wellbeing programs should read the Santa Clara Principles, the Global Network Initiative guidelines, and IWF's annual reports. Then articulate a concrete feature: a mandatory cooldown trigger, a peer-support routing rule, a cumulative-exposure metric that auto-reassigns queues.

Reference the ecosystem. Name-drop the integrations that matter: IWF hash sets, StopNCII for non-consensual intimate imagery, Salesforce for case escalation, NCMEC CyberTipline. Cinder's platform connects all of them. If you've integrated any third-party trust & safety signal (PhotoDNA, PDQ, a custom classifier) detail the latency budget, the fallback logic, the audit trail you kept. Cinder's data shows the company processes 3B+ end-users across clients; they've seen every integration failure mode.

Don't fake the mission fit. The trust & safety labor market flooded after X, Meta, Alphabet, and Amazon cut thousands of roles. Many applicants are refugees chasing any landing spot. Cinder's screeners know the difference between someone who needs a job and someone who believes the work matters. The IWF partnership announcement quotes Cinder: "We want to make it as simple as possible for organisations to strengthen their safeguards." If you can't say why that simplicity is hard-won (why every "simple" workflow hides a thousand policy exceptions) you won't pass the culture filter.

The playbook is narrow by design. Cinder needs engineers who've felt the weight of a wrong call on a child safety case. Who've argued with a product manager about whether reshare friction reduces virality enough to justify the engagement hit. Who've woken up at 3 AM because a classifier drifted and a hate-speech campaign slipped through. If that's not your background, the screen will catch it. If it is, stop listing frameworks. Start telling the stories that prove you've lived the problem.

What Cinder's Process Signals for Trust & Safety Hiring

Cinder's screening gauntlet isn't an outlier — it's the leading edge of a sector-wide recalibration. PwC's 2026 Trust and Safety Outlook, drawn from a survey of 2,000 U.S. adults, frames the shift bluntly: T&S must move "from a reactive operating function to a strategic capability embedded across product design, governance, operations, and ecosystem partnerships." The research identifies AI agents, physical AI, synthetic reality, and machine-speed manipulation as forces expanding the risk surface beyond traditional content moderation. Cinder's demand for engineers who can also map abuse patterns mirrors that mandate: the function now requires people who can build the detection infrastructure and anticipate how adversaries will weaponize it.

The talent market is sending the same signal. Gartner predicts a quarter of candidate profiles could be fake by 2028, and the AI skills gap ranks as the top barrier to enterprise AI integration — education, not workflow redesign, is the primary lever companies are pulling. In construction safety, a parallel labor market, field credibility now outweighs certifications; 70% of the workforce is passive, and the best hires come through relationship networks, not job boards. Cinder's multi-stage screen (technical assessment, post-coding review, cultural alignment) functions as a credibility filter in an environment where polished résumés and rehearsed answers are increasingly cheap to generate.

The hybrid profile Cinder seeks (engineering rigor paired with abuse-pattern empathy) matches the emerging archetype across high-stakes safety domains. Deloitte's State of AI in the Enterprise finds new roles crystallizing: AI operations managers, human-AI interaction specialists, quality stewards. PwC's outlook calls for the future of content moderation to shift "from scaled review capacity to specialized expertise, intelligence, and partner ecosystems." That specialization requires practitioners who understand model behavior and adversary psychology. Cinder's two open roles sit at that intersection.

Regulatory pressure compounds the urgency. UK data shows 75% of women and 52% of Gen Z favor a social-media ban for under-16s; child safety in the age of AI is a named PwC workstream. OSHA enforcement is rising in physical industries; those acts impose systemic risk obligations on platforms. Companies that treat T&S as a cost center staffed by contractors are being forced to rebuild it as a governed, measurable capability. Cinder's salary band reflects that revaluation.

The governance gap is widening. Only one in five companies has a mature model for autonomous AI agent oversight; 58% report at least limited physical AI use today, heading to 80% within 24 months. Agentic AI in T&S operations (evaluator agents, autoraters, AI-led workflows) is arriving faster than the talent to govern it. Cinder's screen, however exacting, is a bet on the only viable pipeline: engineers who have already seen the abuse patterns up close, can write the detection logic, and won't flinch when the adversary adapts. The firms that replicate that filter will staff the function; the rest will keep posting roles that stay open — ghosted by the very people who could do the work.

Working in AI? Zero G Talent tracks the openings: see every open Cinder role, browse AI jobs, the companies hiring, and the people building the field.