Who gets hired, and onto which teams
Protege operates at the intersection of AI model builders and the proprietary data they need to train on, a two-sided platform problem that demands a different hiring profile than a SaaS shop or a research lab. Its board listings show a team already organized around that tension: product and content marketing, forward-deployed engineering split by modality (video, audio), partnerships and business development, and a dedicated technical recruiter to keep the pipeline moving. All six open roles are remote.
The company announced $25 million in Series A funding in August 2025, followed by a $30 million Series A extension led by Andreessen Horowitz in January 2026, marking a capital-intensive growth phase where the platform must onboard data providers and model builders at once. Protege's own site quotes Mahesh Ranganath of Siemens Healthineers' medical imaging division calling the platform an "internal partner… helping us dig into exactly what data we need for the specific problem we're trying to solve, rather than simply being a data catalog." That healthcare use case shows up again in Protege's October 2025 evaluation datasets announcement and its February 2026 collaboration with Vals AI on clinical documentation and medical billing benchmarks, which suggests the forward-deployed engineering roles are not generic integration work but modality-specific data preparation for high-stakes domains.
The functional split on the board maps cleanly to that motion. The Technical Product & Content Marketing Lead sits at the top of the funnel, translating platform capabilities for both sides of the marketplace. The two Forward Deployed Engineer roles (video and audio listed separately) treat data modality as a first-class hiring dimension; a candidate who has built preprocessing pipelines for medical imaging video is not interchangeable with one who has handled multi-channel audio for clinical dictation. The Partnerships Business Development Representative and Strategic Partnerships Manager roles reflect the provider-acquisition side: recruiting hospitals, health systems, and specialized data vendors into the platform. The Senior Technical Recruiter role confirms the company is still in the steep part of the hiring curve.
What the board does not show are the research and core platform engineering teams that must exist beneath the forward-deployed layer, though the public materials only hint at them. Protege's evaluation datasets and benchmarks for healthcare AI, plus its work with Vals AI on clinical documentation and medical billing benchmarks, demand measurement infrastructure, statistical rigor, and domain-specific evaluation design. Those roles either haven't been posted publicly yet or are filled through networks not visible on the job board. The same applies to any core platform engineers building the data ingestion, governance, and privacy controls that the company's positioning implies.
For candidates, the takeaway is structural: Protege hires in modality-specific pods (video, audio, likely text/EMR next) that pair a forward-deployed engineer with a partnerships counterpart, both supported by a product-marketing function that speaks both languages. Backgrounds that map cleanly include ML engineers who have shipped data pipelines in regulated healthcare settings, partnerships leads who have sold into hospital systems or imaging vendors, and product marketers who have translated technical platforms for dual audiences. Remote-first posture means location is not a filter. Domain fluency is.
What it pays
Protege has not published salary bands, and none of the job postings on the board disclose cash compensation or equity ranges. The company's public materials, including funding announcements, product pages, and customer testimonials, are silent on pay. That absence is itself a signal: at this stage, compensation is negotiated case by case, calibrated to the candidate's leverage and the role's strategic weight.
What we do know comes from the funding timeline. A $25 million Series A closed in August 2025; a $30 million Series A extension led by Andreessen Horowitz followed in January 2026.
The roles themselves clarify where the money flows. Forward Deployed Engineer, Video and Forward Deployed Engineer, Audio sit at the customer interface, shipping integrations and debugging model-data fit in production.
Equity mechanics at a Delaware C-corp at this stage typically follow the standard template: four-year vesting, one-year cliff, monthly thereafter. The compliance-heavy culture implied by the "Privacy by design" and "Responsible AI" positioning usually means clean cap tables, no unusual liquidation preferences, and standard double-trigger acceleration on change of control, but without a published equity FAQ or offer letter template, those are inferences, not guarantees.
Remote-first hiring adds a geographic variable. All six board listings carry the "Remote" tag. If Protege follows the prevailing Series A/B remote model (GitLab, Linear, Sourcegraph), cash bands are national with no location differential, while equity grants are flat regardless of zip code. That helps candidates in lower-cost metros but compresses the spread between a Bay Area hire and one in Austin or Raleigh.
Bottom line: candidates should enter conversations with a clear number in mind, anchored to public comp data for comparable roles at Series A/B AI infrastructure companies (Scale AI, Weights & Biases, Hugging Face, Pinecone), and prepared to negotiate the equity-cash tradeoff explicitly. The company has the capital to pay. It has not chosen to publish the menu.
How the hiring process works
Protege runs a consistent five-stage interview journey regardless of function, whether sales, product, operations, or engineering, with only the technical assessment varying by domain. The structure mirrors patterns candidates encounter at comparable high-growth startups: an informal introductory call, a screening call, a culture-fit conversation, a skills or technical round, and a final decision meeting with a senior leader.
The first call is explicitly informal. Recruiters use it to answer candidate questions about the process and the role, and to gauge baseline alignment before investing deeper time. The screening call is where recruiters test whether the candidate has done the work: "why this company?" is a standard question, designed to surface whether the applicant understands the business model, the market, and the trajectory. Recruiters also probe the five-to-ten-year plan to assess ambition and whether the candidate's growth curve matches the company's. Candidates who frame the move as "just a job" or cannot articulate how their skill development compounds with the organization's roadmap tend to stall here.
Round three evaluates culture fit through behavioral questions mapped to core competencies: leadership and initiative, problem-solving and analytical thinking, teamwork and collaboration, pressure handling, conflict resolution, innovation, customer focus, adaptability, communication, and achievement orientation. Candidates who prepare STAR+RC stories (Situation, Task, Action, Result + Reflection/Change) that each hit multiple competencies consistently report better outcomes than those who improvise. Recruiters say mapping stories to these dimensions shifts the conversation from defensive justification to evidence-based dialogue.
Round four is the domain-specific skills test. For sales roles this means a live pitch or mock discovery call; for technical roles, a coding or architecture exercise; for product, a case study. This round resists last-minute cramming. Candidates who rehearse a tight 10-minute run-through of their approach perform markedly better than those who wing it. A failed skills round can trigger a confidence spiral; recruiters describe candidates scrambling for external data mid-call and inadvertently signaling incompetence.
The final round with a senior leader is less about technical verification than strategic alignment and closing. Candidates who close with high-impact questions, about team priorities, board-level OKRs, or the interviewer's own career path, leave a stronger signal than those who ask generic culture questions.
Recruiter screening criteria prioritize demonstrable research on mission, values, recent news, product suite, and org structure (including the hiring manager's LinkedIn activity); vocabulary fluency in the top five to seven skills for the target role; and a narrative that connects past impact to future trajectory. Common disqualifiers include inability to name a specific company initiative or metric, generic answers to "why us?", STAR stories that lack quantified results or reflection, and treating the process as a test to pass rather than a mutual evaluation.
Preparation tactics that consistently appear in successful accounts: asking the recruiter in advance who will attend each round and what format to expect; connecting with the interviewer on LinkedIn beforehand; networking with current employees in the same function to harvest role-specific vocabulary; and rehearsing opening and closing statements as scripted three-liners. The system rewards candidates who operate as if they are already inside, framing answers from the perspective of a contributor solving the team's problems, not an applicant seeking validation.
Where the work happens
Every role listed on Protege's careers page and on this board (Forward Deployed Engineers for video and audio, a Technical Product & Content Marketing Lead, Partnerships and Strategic Partnerships managers, a Senior Technical Recruiter) carries the "Remote" designation. There is no headquarters campus, no integration floor, no machine shop, no test cell. The funding rounds financed product development, data licensing, and go-to-market motion, not brick-and-mortar infrastructure.
That absence is the story. Protege's product is a data platform that connects AI companies with proprietary, non-public datasets across healthcare, media, audio, and motion capture. The work is API integration, data pipeline engineering, contract negotiation, and customer onboarding, all of which run on laptops and cloud tenants. Siemens Healthineers' Mahesh Ranganath described Protege as "an internal partner… helping us dig into exactly what data we need for the specific problem we're trying to solve." That partnership happens over shared notebooks, video calls, and secure data rooms, not in a hallway conversation outside a cleanroom.
The Stanford Protégé project, the open-source OWL ontology editor developed at Stanford Medicine, does have a physical presence: a short course held June 23–25, 2026, at Stanford University. But that is a separate entity. The Stanford tool is a desktop and web application for building biomedical ontologies (ICD-11, the NCI Thesaurus, OBO Foundry ontologies), written in Java with a plug-in architecture, and used by researchers, governments, and the WHO. Protege the company shares a name and a domain adjacency: both serve AI and data workflows, but the two are distinct. Conflating them would mislead candidates.
LexisNexis Protégé adds a third layer: a legal AI assistant embedded across Lexis+®, Lexis Create+, Nexis+™, Lexis Machina®, CounselLink+™, CourtLink®, Intelligize+ AI™, and PatentSight+™. It runs inside Microsoft 365, including Word, Excel, PowerPoint, Teams, and Outlook, and connects to iManage, SharePoint, OpenText, and Google Drive. Its workspace is the document pane a litigator already has open.
For a candidate evaluating Protege, the practical takeaway is simple: you will not commute to a lab. Your integration floor is a CI/CD pipeline; your test cell is a staging environment spinning up against a customer's data sample. Collaboration is asynchronous by default. Written updates, recorded demos, and shared Notion pages form the backbone, with synchronous windows scheduled across time zones. The company's blog posts come from distributed contributors.
This model has trade-offs. Deep technical debugging on a customer's on-prem environment may require a site visit, but those are exceptions scheduled per engagement, not a standing lab capability. Onboarding relies on documented runbooks and pair-programming sessions rather than a shadowing week at a physical desk. The Senior Technical Recruiter role itself is remote, signaling that even the hiring function operates without a central office.
Candidates who thrive in a distributed, write-heavy, API-first workflow will find the environment matches the product: secure, data-centric, and location-agnostic. The only "site" that matters is the customer's data estate, and Protege's job is to reach it from wherever the engineer sits.
What kind of person thrives here
Protege's published job listings, customer accounts, and the way the company describes its own work converge on a fairly specific profile. The same traits surface whether you read a Forward Deployed Engineer posting, a Strategic Partnerships Manager brief, or a quote from a customer talking about what working with Protege actually feels like.
The strongest signal is customer-obsession with a technical spine. The Ranganath quote, treating customer problems as your own and then digging until the data or integration is right, captures the disposition Protege's hiring pages reward. Listing video and audio as separate specialties in the Forward Deployed Engineer roles suggests Protege splits hires by domain fluency rather than recruiting generalists and hoping they ramp.
Cross-functional communication shows up everywhere in the public record. Sean Fitzpatrick, CEO of LexisNexis North America, UK, & Ireland, said Protégé General AI responds to "customer requests for secure access to general-purpose models while maintaining control over AI behavior within legal workflows." The customer preview program, which the company reports has drawn roughly 200 law firms, corporate legal departments, and law schools, runs as a multi-stakeholder sales and product feedback loop. Anyone thriving in that environment has to translate between model researchers, enterprise buyers, and compliance teams in the same meeting.
Domain judgment matters because Protege sits on regulated, high-stakes data. The healthcare benchmarks launched in October 2025 and the February 2026 Vals AI expansion into clinical documentation and medical billing put the company inside workflows where a wrong label has real consequences. The privacy-by-design framing reads less like marketing than like a hiring filter. People who treat data as something they have to defend, not just ship, are the ones likely to last.
The company also rewards builders who can ship in public. The recent board postings for a Technical Product & Content Marketing Lead and a Senior Technical Recruiter sit alongside the engineering and partnerships roles, which is a clue that Protege expects technical staff to write, recruit, and sell their own work, not hand it off.
The simplest test, then, is the customer's: can you describe, in one paragraph, a specific problem a customer handed you and the exact data or system you built to solve it? Protege's hiring bar reads less like a list of credentials than like that paragraph, a proof that you already work the way the company needs you to.
Working in frontier tech? Zero G Talent tracks the openings: see every open Protege role, browse frontier tech jobs, the companies hiring, and the people building the field.