Infinitus Systems is building a team around product, interface, engineering, operations and people work, with San Francisco as headquarters and a remote-work program. Eight recent board postings show distinct hiring needs across agent behavior, software engineering, data, customer delivery, outside resources and workforce support.
Infinitus’s own engineering hiring write-up says candidates should demonstrate how they will perform on the job rather than just how quickly they can solve an algorithm puzzle. For a healthcare agent company whose automated agents hand exceptions to people, that standard fits the operating model: product decisions have to hold up when software, clinical judgment and human review meet.
The postings do not target one generic “AI” profile. They call for product managers who can define agent behavior, engineers who work across interfaces and conversation systems, a resource manager to coordinate outside inputs, customer-delivery leadership and people support for a fast-growing AI healthcare company. The roles differ, but the hiring evidence points to candidates who can connect product choices to healthcare workflow behavior and human review.
Who Gets Hired and Onto Which Teams
The recent board makes Infinitus’s team structure unusually visible. Product work occupies several distinct areas, while engineering, data, customer delivery, resource coordination and human resources sit around it. The roles below are not interchangeable versions of the same job:
| Opening | Function | Hiring focus |
|---|---|---|
| Product Manager, Payor Agents | Product | Agent workflows for payor calls and healthcare payments |
| Product Manager, AI Hub | Product | Shared AI capabilities with human review and escalation |
| Workflows and Memory - AI PM | Product | Agent workflows, context, entities, knowledge and memory |
| AI Full Stack Engineer | Engineering | Full-stack software delivery |
| Data Analyst | Data | Actionable, company-wide data |
| Customer Success Manager | Customer operations | Customer support and delivery |
| Vendor & Resource Manager | Operations | Vendor and resource coordination |
| HR Business Leader | People | Performance, employee relations, compensation and talent planning |
Infinitus’s careers page exposes filters for Customer Success, Engineering, Operations, People and Product, while its profiles include Customer Engineering, Marketing and Partnerships. The board adds data analysis, vendor and resource management, and an HR business role. Design does not appear as a separate line in the openings, but the company’s career profiles place interface, customer-data and conversation work close to product and engineering. A medical-lead profile also shows that clinical judgment contributes to the company’s work even when it is not a distinct job family.
The product descriptions make the company’s operating expectations unusually concrete.
The product bar: own the whole change
In the product-role brief, Infinitus said:
“Take a meaningful slice of Studio from idea to shipped, owning discovery, requirements, and launch for the areas closest to you.”
That mandate covers discovery, requirements and launch. The fuller role description also names customer discovery, shipping, measuring, iterating and balancing performance, latency, cost and healthcare requirements. A roadmap is useful only if the person who owns it can also explain how the work behaves in practice.
The relevant evidence is how delivery connects a user’s problem to system behavior and the step that keeps it working. The same chain continues through interface and operations, where a product decision meets a customer’s reality.
Design, operations and the human handoff
Interface work appears throughout Infinitus’s career profiles. One engineering profile focuses on frontend components of the Infinitus stack and connects that work to the customer-portal launch and customer-data segmentation; another covers the natural-language-understanding systems behind its digital assistant, from rule-based systems to self-learning models. Those descriptions place interface and conversation work inside system behavior rather than treating it as a final coat of polish.
Product candidates need enough technical and interface fluency to ask useful questions, not enough to impersonate engineers. They should know what the interface exposes, what the model handles reliably and where a person must take over. Infinitus’s “Meet the Infinitus human in the loop” profile explains that humans play key roles in delivering the customer experience and that modern AI systems learn from human feedback. That makes human review and handoff part of the operating context.
In its careers copy, Infinitus said:
The care exists. The communication and coordination doesn’t – not yet, not at the scale it needs to.
That statement frames the operating work behind the openings: customer leaders, resource planners and people managers who can turn product use into a repeatable service, preserve context and route exceptions. For these roles, evidence of coordination matters alongside a polished résumé.
The company’s AI-first, human-backed approach and its July 2026 AI-first hub announcement make the division explicit: automated agents handle routine requests, while trained experts take exceptions, escalations and cases involving clinical nuance or ambiguity. The agent passes context into the review. Product, interface and operations are therefore not adjacent specialties; they define the same service boundary.
What the right candidate proves
Ankit Jain said:
Models and interfaces will change. Entire technology stacks will disappear. But patient needs will remain the same.
Jain’s warning gives applicants a practical filter. Technical novelty tells a recruiter little about fit. The stronger application starts with a patient or customer problem, identifies a failure in the current journey and coordinates product, interface, engineering and operating choices to fix it.
The evidence can come from product, interface, operations or people work, but it must show a real transition between teams rather than a list of credentials. Product candidates show a shipped workflow; design candidates explain an interface and service transition; operations candidates diagnose and fix a coordination failure; people leaders show how they changed the organization.
Build the application around one concrete case. Skip the tool inventory. Show the user need, the product and interface decision, the operating bottleneck, the handoff between teams and the result. Rewrite one résumé bullet in that order before applying. If the example omits what shipped or the coordination around it, it will not support your case for these roles.
That application is the part candidates control. The terms of the offer are not, and the public record has a conspicuous gap.
What It Pays
Infinitus can be specific about the work while remaining silent about its price.
The posted roles and the missing numbers
The first-party job board provides role and location details but no base-salary minimum or maximum for any of its entries. The omission is uniform rather than limited to one role. Without a range, a candidate cannot make a factual comparison with another AI employer; an estimated salary would turn missing data into false precision.
The openings reveal hiring priorities, not a compensation hierarchy. Product, engineering, customer and resource work can carry different pay, and the board gives candidates no way to compare them. The location labels also do not explain the policy: candidates cannot tell whether a role’s level or location changes its band.
Infinitus’s company benefits listing includes company equity, performance bonuses, health insurance, a 401(k), parental leave, paid time off and a remote-work program. The listing describes parental leave and paid time off as “generous” but gives no number of weeks or days.
The record leaves key terms undefined. It gives no bonus target, performance conditions or guarantee, and no form, value, vesting schedule or refresh policy for equity. It also omits the 401(k) match, health-plan premiums and deductibles. “Company equity” can carry very different value depending on those terms, so candidates should read the grant details before treating it as cash.
In its November 10, 2021 financing announcement, Infinitus said it had closed a $30 million Series B and brought total funding to $51.4 million. The company planned to invest in talent, expand operations and continue product delivery. The announcement shows its 2021 priorities; it does not set today’s salary or hiring budget.
Ask the recruiter for the numbers. Request the base range for the role, level and location; the bonus target and performance criteria; the equity instrument, grant terms and vesting schedule; and the geographic pay policy. Use those written terms to compare competing offers rather than the company name, funding history or a benefits checklist.
Those terms settle the offer. They do not settle whether the candidate can do the work. Public hiring evidence begins with timing.
How the Hiring Process Works and What Gets Candidates Through It
Public interviews offer a rough hiring-time benchmark, but they do not map the full process. Infinitus has published an engineering interview model involving code review, mock-outage triage and system design, but the public evidence does not establish one company-wide sequence, take-home assignment or fixed number of rounds.
Glassdoor’s user submissions provide the following benchmarks:
| Role category | Reported time to hire | Evidence base |
|---|---|---|
| All job titles | About 3 weeks, or 21 days | 20 user-submitted interviews |
| Member of Technical Team | About 2 months, or 60 days | 1 user-submitted interview |
The technical estimate rests on one account and should be treated as an anecdote, not a company norm. Glassdoor users rated the interview experience 50% positive and its difficulty 2.7 out of 5. Applicants for Machine Learning Engineer and Data Engineer roles reported the hardest interviews. The ratings do not identify the stages, while Infinitus’s engineering hiring write-up gives a firmer signal: candidates are being asked to show job-relevant judgment rather than puzzle speed.
In a Pulse 2.0 interview, Jain said Infinitus’s platform runs more than 100 models, from small systems designed to reduce latency to large language models, with human-in-the-loop machine learning before, during and after calls. Technical candidates should prepare to defend how a system behaves in production rather than rely on model demos.
What a recruiter screen can and cannot tell you
The job postings and other public material do not confirm a recruiter screen, its questions or whether every role starts with one. Anyone who describes a universal interview loop is guessing.
If a recruiter contacts you, turn the conversation into a two-way fit check. Prepare a short account of relevant work, including one result and one failure, and connect it to the role’s operating problem. Broad claims about working with AI will carry little weight.
This matters most for product and operations candidates. Healthcare buyers weigh return on investment, build trust over time and seldom ship software on a consumer app’s schedule. A candidate who can map a complex process, handle exceptions, assign owners and measure an outcome offers stronger evidence than broad job titles.
The role descriptions provide firmer signals than interview ritual.
The application signals tied to the work
The Payor Agents and AI Hub openings point to product judgment around autonomous phone workflows and shared AI features.
A credible application starts with a messy process, not a polished model. In that Pulse 2.0 conversation, Jain said Infinitus’s agents automate payor calls from beginning to end, move through interactive voice-response menus, collect required information and push back when a payor representative gives incorrect details. Product candidates should explain how they defined product behavior, set boundaries, handled bad inputs and judged whether automation improved the workflow.
In a Fierce Pharma Week interview, the transcript records:
So a safe agent, a trusted agent is one that you know where it's coming where it's coming from.
The same interview describes trusted agents as needing clinical and administrative guardrails, knowledge-based responses and natural conversation. Applicants should be able to explain where automation should stop, how a system should admit uncertainty or when to bring in a person.
The Workflows and Memory opening covers agent workflows, context, entities, knowledge and memory. In the same conversation, Jain said the longest call Infinitus processed that year lasted 3 hours 15 minutes and contained 306 conversational turns. Candidates should explain state, memory, reliability and how they would detect drift in a long-running workflow.
The Vendor & Resource Manager title points to vendor and resource coordination. The HR Business Leader description covers performance, employee relations, organizational effectiveness, compensation, talent planning, workforce decisions, scalable People processes, compliance and immigration matters. Infinitus has not published a checklist, but the role titles and descriptions point to coordination, ownership and process control.
What will not clear the bar
Infinitus has published no formal disqualifiers, and no evidence supports a fixed rejection rule. Its hiring evidence nevertheless suggests several poor-fit signals:
- Promise quick launches but leave healthcare workflows, compliance or implementation constraints unaddressed.
- Focus on model size or novelty while avoiding questions about latency, accuracy, failure handling or human review.
- Seek broad autonomy without guardrails, attribution or a path to a person who can resolve an unknown question.
- Present general AI credentials without evidence of operational or product judgment.
These signals flag fit; they are not instant-rejection rules. A candidate can still advance by explaining the reasoning behind a decision. One who cannot may struggle to explain the work.
Rehearse one example first in a concise recruiter introduction and then in a detailed product, operations or technical discussion. A coherent answer should not change with the format. Before the calendar becomes concrete, read the posting’s location label: San Francisco Office or Remote.
Where the Work Happens
Infinitus is headquartered in San Francisco and recruits for both headquarters-linked and remote work. Its current board divides the openings by location:
The following table summarizes the distribution of recent job openings by their work-location label, pairing the specific opening counts with their corresponding proportional share of the total board.
| Work-location label | Recent openings | Share of the board |
|---|---|---|
| San Francisco Office | 6 | Three in four |
| Remote | 2 | One in four |
The mix describes the recent recruiting represented by the board, not the workforce. A location tag is an advertising convention, not a census of where employees live, how often they travel or how many work from headquarters. The board cannot support turning six office labels into a mostly-office workforce any more than it can establish a universal home-office policy.
The assignment is role-specific. The resource-management and customer-success openings carry remote labels, while the other roles on the board attach to San Francisco. That pattern indicates how Infinitus is advertising particular jobs, not a blanket rule applied after hire.
The remote customer role is not tied to one local market. A 2023 Infinitus announcement described customers across independent ambulatory surgery centers, specialty pharmacies, national payors and pharmaceutical patient-access programs. The posting does not say where the employee lives, whether the role expects travel or how often the employee works with headquarters.
Built In labels several recent postings as hybrid, but neither it nor the first-party board specifies a weekly schedule, required office days or a universal home-office policy. Read each first-party label literally until the recruiter supplies details.
An office-directory record from late October 2026 lists Infinitus’s headquarters at 450 Townsend St, Suite 500. It identifies San Francisco as the operating center for corporate functions, major leadership and technology development. That supports a company with centralized direction even when the work is distributed. The record offers no floor plan, desk layout, amenities or headcount, so it cannot establish daily office culture.
The location label does not change the product’s operating model. Infinitus’s July 2026 announcement says routine requests go to AI agents, while trained experts handle exceptions, escalations and clinical or ambiguous cases, with context passed at handoff. The board does not disclose whether remote employees work with the San Francisco office or how often they travel.
For applicants, the distinction is practical. A “San Francisco Office” posting connects the role to the headquarters where corporate functions, leadership and technology development sit. A “Remote” posting states that the role is remote but does not disclose the detailed arrangement. Before an on-site interview or offer, ask which office days or events matter, which internal team you would support and how that team overlaps with San Francisco. Neither the address nor the role tag answers those questions.
The address identifies the center; the role tag shows how you enter it.
Location tells you where the work sits; judgment tells you whether it works.
Who Thrives Here
The people who thrive at Infinitus Systems keep their judgment precise when inputs are messy and their work useful when metrics look good. Infinitus’s record calls for applied judgment: candidates must understand a healthcare workflow, define what must be right, recognize uncertainty and involve a person when the stakes justify it. The evidence reaches beyond engineering into product, interface, operations and people work, but it describes engineering in greatest detail.
Real work beats interview theater
In an October 30, 2025 engineering post, Infinitus said:
“Candidates want to demonstrate how they’ll perform on the job, not just how quickly they can invert a binary tree.”
The post made a practical argument about hiring evidence. LeetCode-style questions can test pattern recognition and persistence, but they rarely reproduce the noise, dependencies or judgment calls of daily work. Candidates should define system behavior, handle imperfect information and explain tradeoffs. Product and operations candidates need the same evidence as engineers: a clear approach to a realistic constraint.
Precision and calibrated judgment
Infinitus’s engineering team says its voice agents can gather scores of fields from one payor call, including plan details, network status, patient cost-share and prior-authorization requirements. A vague or wrong answer can change a patient’s benefit, cost or authorization. The same write-up says its scoring guidance tells models to abstain on critical fields rather than guess, while allowing variation in supporting data that downstream systems can normalize. People who thrive there learn the same discrimination: identify the missing or malformed detail that could change an outcome, and recognize when a less exact field can pass through safely.
That standard runs through both model scoring and human review. The model should pass uncertainty forward; the reviewer should know what to trust and when to intervene.
The escalation process makes that judgment visible. Infinitus says a flagged call is immediately escalated to a human expert for review and that it can flag tasks needing additional customer review in as little as 15 minutes after a call completes. Admitting uncertainty becomes operational competence when the speaker can identify it and route it to the right person.
Production thinking beats a good demo
A March 18, 2026 engineering post separated a model’s chance of succeeding in at least one attempt from its chance of succeeding across repeated attempts. Potential performance is not operational reliability. A production system cannot depend on a lucky retry; it must behave consistently on the inputs customers actually send.
The cost reaches the people around the system. Infinitus said a stream of false positives can tire reviewers, weaken trust and lead teams to dismiss a real anomaly. Product and operations candidates should explain what the team monitors, who responds to an alert and how a failure reaches the customer before a dashboard looks healthy. A polished demo does not prove a dependable service.
Product and operations judgment
Infinitus’s product write-up explains its operating demands more concretely: output extraction turns messy, unstructured inputs into fields that downstream systems can use, and in some cases customers receive results within half an hour.
This work demands more than feature ideas. Candidates who define field behavior, downstream dependencies and exception handling offer stronger evidence than those who promise speed through vague automation. Design needs the same discipline, though the available product post gives little design-specific detail. The narrower conclusion favors people who connect interface decisions to the wider workflow and its failure paths over those who offer isolated interface polish.
The Vendor & Resource Manager and Customer Success Manager titles place resource coordination and customer delivery alongside that product work. In a team that mixes San Francisco office-linked and remote roles, written clarity and early escalation matter. A blocker left in someone’s notebook cannot wait for a hallway conversation.
Openness with accountability
Undated Glassdoor summaries describe a relaxed atmosphere with supportive teams and open-door policies. Other reviews say the culture has become more metrics-driven, with a possible effect on morale. Infinitus appears to value support and measurable output at the same time.
The people likely to thrive can ask for help early without waiting to be rescued and accept direct feedback without treating one number as a verdict on their worth. They use metrics without treating them as the only goal. Infinitus’s own discussion of false alarms shows why: a metric detached from the decision it should support creates more work and less trust.
The people-leadership opening adds another operational dimension: the role covers performance and compensation programs, employee relations, talent planning and workforce decisions. Taken with employee reports about a more metrics-driven culture, it suggests candidates should know what the numbers mean rather than simply produce better reports.
Before applying, test your fit against one Infinitus workflow. Map its critical outputs, acceptable variations, human-escalation point and failure signal. Show what happens when the data is incomplete or wrong.
A call can run for hundreds of turns without a dramatic break. It can go wrong one unclear field at a time. The strongest application shows where the system stops, who takes over and what happens next.
Working in frontier tech? Zero G Talent tracks the openings: see every open Infinitus Systems role, browse frontier tech jobs, the companies hiring, and the people building the field.