How to Apply to Arize

Key Takeaways

Compensation at Arize is competitive with other Bay Area AI infrastructure startups at a similar funding stage, which means it is structured to be meaningfully better than what large-cap technology companies offer when measured on total expected value, including equity. Cash compensation for engineering roles in the United States typically ranges from approximately one hundred fifty thousand dollars at the early-career level to roughly two hundred forty thousand dollars at staff and principal levels, with senior engineers and tech leads commonly landing in the one hundred ninety to two hundred twenty thousand dollar range depending on location and specialty. Applied AI and research-leaning roles often carry a premium. New York compensation tracks closely with Bay Area bands. Bangalore compensation is benchmarked against the top tier of the Indian market for ML platform talent, which puts Arize at or above the upper range of what competing AI startups and large multinationals pay there. Equity is granted as stock options on a four-year vesting schedule with a one-year cliff, and the size of grants for engineering hires has historically been meaningful enough that early and mid-stage employees have a real stake in the outcome of the company. Specific grant ranges are role-dependent and are best discussed candidly with the recruiter once you have advanced into the loop. Benefits in the United States include comprehensive medical, dental, and vision coverage with the company covering a substantial share of premiums, a 401(k) plan, generous parental leave, a flexible paid time off policy, a home office stipend for remote and hybrid employees, learning and development budgets, and reimbursement for relevant conferences and books. The company observes a hybrid model for most teams in Berkeley and New York, with two to three days a week in office considered the norm for most roles, and offers fully remote positions selectively when the role and candidate profile justify it. Bangalore operates on a comparable hybrid arrangement. Candidates negotiating offers should focus less on incremental cash improvements and more on equity composition, refresh expectations, and the seniority leveling, since the long-term value of an Arize offer is concentrated in the equity package and in the level at which you enter, which influences the trajectory of subsequent promotions and refreshes.

About Arize

Arize AI sits at the center of one of the most consequential shifts in enterprise software: the move from deploying machine learning models as research artifacts to running them as load-bearing production systems whose failures must be detected, diagnosed, and corrected in real time. Founded in Berkeley in 2020 by Aparna Dhinakaran (CEO and Chief Product Officer) and Jason Lopatecki (Executive Chairman), the company emerged from a thesis its founders had spent years developing inside their previous employers, including TubeMogul and Adobe, where Lopatecki had built ad-tech platforms that ran billions of model predictions daily and Dhinakaran had led ML engineering. They saw the same pattern repeatedly: teams could ship models to production, but no one could tell when those models began to drift, hallucinate, or quietly degrade against silent failure modes that traditional application monitoring tools were never designed to catch. Arize was built to be the system of record for that problem, and roughly 150 people now work across its Berkeley headquarters, its New York office, and its engineering hub in Bangalore to extend that mission across the rapidly expanding surface area of generative AI. Hiring at Arize is highly intentional. The company is not running a hypergrowth headcount sprint; it is running a craft-and-conviction filter where every engineering, research, and go-to-market hire is expected to take ownership of an ambiguous problem and ship a measurable result against it. This guide explains how that filter actually works, what each part of the funnel is genuinely testing, and how to position yourself credibly whether you are an applied ML engineer who has been instrumenting LLM agents for the past year or a platform engineer who has never touched an evaluation pipeline but understands distributed systems at depth. We have written it specifically for candidates who want to move past surface-level company research and arrive at an interview with Arize understanding the substance of what the company sells, who it sells against, and what kinds of problems its teams are actually solving inside the codebase right now.

Application Process

1
The Arize hiring process is structured to filter for substance over polish, and
The Arize hiring process is structured to filter for substance over polish, and it typically takes three to five weeks from application to offer for engineering and research roles. The process begins with an application submitted through the company's Greenhouse careers portal, which is the single source of truth for active openings. Recruiters and hiring managers triage inbound applications quickly, often within a week, and they read resumes for evidence of three things in particular: ownership of shipped work that touched production, relevant exposure to ML systems or LLM applications, and the ability to write or communicate about technical problems in a way that suggests you can collaborate with research-leaning colleagues. The first conversation is a thirty-minute call with a recruiter that focuses on your motivations, your understanding of what Arize does, and the basic logistics around compensation expectations, work authorization, and location. Candidates who can articulate why they want to work on observability and evaluation specifically, rather than generic interest in AI, advance reliably. The second stage is a hiring-manager screen, typically forty-five to sixty minutes, where the manager probes the depth of your most relevant project. Expect specific, pointed questions about architecture decisions, trade-offs you rejected, and what you would do differently. Vague answers do not survive this stage. The technical loop that follows depends on the role. Software engineers and platform engineers go through a coding interview that emphasizes practical problem solving in a real editor with the language of your choice, followed by a system design interview centered on a problem in Arize's domain, such as designing a high-throughput trace ingestion pipeline, an evaluator scheduling system, or a multi-tenant query layer for embedding analysis. AI engineers and applied scientists face an evaluations or LLM-systems interview where you might be asked to design an offline evaluation harness for a retrieval-augmented generation pipeline, critique a flawed prompt evaluation strategy, or walk through how you would detect agent drift in production. Solutions engineers and forward-deployed engineers complete a customer-scenario exercise, often involving a take-home or live walkthrough of how you would help a hypothetical customer instrument a complex LLM application and surface the right failure modes. The final stage is an onsite or virtual onsite of three to five interviews that includes additional technical depth, cross-functional collaboration with product or research, and a values conversation with a senior leader, frequently Aparna or another executive. Offers are typically extended within a week of the final round, and Arize moves quickly when it has conviction. References are taken seriously and back-channel checks are common, especially for senior roles, so candidates should ensure their stated work history is accurate and that the colleagues they list as references can speak in detail to specific shipped outcomes.

Resume Tips for Arize

recommended

If you are serious about an offer from Arize, the highest-leverage preparation y

If you are serious about an offer from Arize, the highest-leverage preparation you can do is to install Phoenix locally, instrument a small LLM application of your own, and form an opinion about what works well, what is missing, and what you would build next. Interviewers across multiple rounds will probe whether you have engaged with the product, and candidates who can speak from direct experience stand out from those who have only read marketing pages. Read the Arize engineering blog and the Phoenix documentation in depth, paying particular attention to recent posts on agent evaluation, tracing patterns, and evaluator design, because these reflect what the company is currently investing in and will frequently appear in interview discussions. Watch recent talks by Aparna and other Arize technical leaders from conferences like AI Engineer Summit, MLOps World, and Ray Summit, where they discuss the company's roadmap and technical philosophy candidly. When you discuss your own work, lead with specifics: what the system did, what it was measured on, what you owned, what you would do differently. Avoid abstract claims about scale or impact that you cannot defend with numbers. In system design rounds, do not jump to a solution. Spend the first several minutes clarifying requirements, naming the constraints you are designing against, and stating your assumptions explicitly. Arize interviewers value structured thinking and are comfortable with candidates who change their mind partway through a problem when new information emerges. In the values and behavioral round, prepare specific stories about times you disagreed with a colleague and how you resolved it, times a project you led missed its goal and what you learned, and times you mentored or were mentored by someone in a way that changed your work. The company is pattern-matching for self-awareness and growth orientation, not for polished narratives. Apply through the Greenhouse portal directly rather than through aggregators, because the company's recruiting team triages its own pipeline and applications submitted through the official channel are tracked properly. If you can secure a warm introduction through a current employee, do so, but do not wait on a referral if you do not have one; recruiters read every application that comes in, and the bar for getting into the funnel is lower than candidates often assume. The bar for getting an offer, however, is high, and the candidates who clear it are the ones who treat the interview process as a real test of fit rather than a performance to be hacked.

ATS System: greenhouse

Apply via boards.greenhouse.io/arizeai.

Use exact Greenhouse keywords from posting
Highlight ML platform / observability experience
Show LLM evals familiarity (Phoenix open-source)

Interview Culture

Arize's internal culture has three load-bearing characteristics that shape day-to-day work.

The first is intellectual seriousness without academic pretension. Engineers are expected to engage with hard technical material, including current research, but the bar for any decision is whether it ships a better product to customers, not whether it sounds sophisticated in a planning document. The second is operating leverage through shared infrastructure. The company invests heavily in internal tooling, in the open-source Phoenix codebase as a forcing function for clean abstractions, and in documentation that allows new engineers to ramp without bottlenecking on tribal knowledge. Engineers who like to invest in their own productivity and the productivity of their colleagues do well here. The third is customer obsession in a literal, operational sense. Engineering and product teams sit close to the customer success and forward-deployed engineering teams, customer problems are routinely discussed in internal Slack channels and engineering syncs, and feature work is regularly traced back to specific named accounts whose workflows it improves. This is a company where shipping a feature that no customer asked for or uses is treated as a problem to be examined, not celebrated. The pace of work is high but not abusive. There are sprints around major launches, including conference moments and large enterprise deployments, where hours expand. Outside those windows, the expectation is sustained, focused work rather than performative overwork. The company is led by founders who are visible, accessible, and demanding in a way that sets the cultural tone: Aparna is widely respected by employees for her combination of product depth and direct communication, and Jason brings an operator's lens to scaling questions. New hires consistently report that the most surprising thing about Arize is how flat the company feels for its size, with senior engineers and individual contributors regularly working directly with the founders on consequential technical decisions. The company is also deliberate about diversity and inclusion in a way that goes beyond statements: hiring panels are constructed to include diverse perspectives, and the company has invested in pipelines that bring underrepresented engineers into ML infrastructure roles where the broader industry has historically been narrow.

What Arize Looks For

Arize hires for a specific intersection of traits that the company calls, internally and informally, the builder-researcher-collaborator profile. The builder dimension is the most heavily weighted: the company wants people who have shipped real systems to real users, who can point to code that is currently in production, and who have learned the discipline of operating that code through incidents and iteration. Pure research credentials without shipping experience are a poor fit for most engineering roles, even on the applied science teams, because Arize's products are infrastructure that other engineers depend on, not papers. The researcher dimension reflects the reality that Arize's domain is moving faster than any framework or playbook can keep up with. Successful hires read papers in arxiv categories like cs.CL and cs.LG with regularity, can credibly discuss recent work on LLM evaluation methodology, agent benchmarks, retrieval evaluation, and reasoning failures, and can translate that literature into product decisions. You do not need a PhD, and most engineers at Arize do not have one, but you do need genuine curiosity about where the field is going and the discipline to keep up. The collaborator dimension is where Arize's culture is most distinctive. The company runs on tight cross-functional pods that pair engineering, product, and research closely, and it values people who can disagree productively, write clearly, and bring colleagues along to a shared technical position. Hiring managers look for candidates who can describe past collaborations specifically, who give credit to teammates without prompting, and who can articulate what they learned from colleagues who pushed back on their initial proposals. The skills that compound most heavily across roles are strong Python proficiency, working knowledge of the OpenTelemetry data model and distributed tracing concepts, comfort with TypeScript and React for front-end-adjacent work, hands-on experience with at least one LLM provider API at scale, and exposure to evaluation frameworks for either traditional ML or LLM systems. For platform and infrastructure roles, depth in distributed systems, columnar storage, streaming ingestion, and high-throughput query engines is highly valued. For AI engineering and applied research roles, prior work on RAG systems, agentic workflows, fine-tuning, or evaluation harnesses is a meaningful differentiator. Arize is honest with itself about the kinds of candidates who do not thrive: people who treat the role as a stepping stone, people who optimize for individual visibility over team outcomes, and people who cannot operate in ambiguity without an existing playbook. If you need a fully scoped problem handed to you, this is not your company.

Frequently Asked Questions

Is Arize AI hiring fully remote engineers in 2026?

Arize is selectively hiring remote engineers in the United States for specific roles, but the default expectation for most engineering positions is hybrid attendance at the Berkeley or New York offices two to three days per week. Bangalore-based roles operate on a similar hybrid pattern. Remote eligibility is decided per role and is typically reserved for senior or specialist hires whose skills are difficult to source locally. Always check the location field on the Greenhouse posting and confirm with the recruiter during the first call rather than assuming based on the title.

What is the difference between Arize Phoenix and Arize AX, and which one will I work on?

Phoenix is the open-source LLM tracing and evaluation library distributed under a permissive license and used by individual developers and teams to instrument applications locally or in any environment. Arize AX is the commercial enterprise platform that adds security, multi-tenancy, datasets, experiments, online evaluators, agent monitoring, classical ML observability, and the analytics layer that customers license. Many engineering teams contribute to both because the products share core abstractions, but role descriptions usually indicate the primary surface area. Open-source contributions to Phoenix are highly visible across the company and are a meaningful part of the engineering culture.

How much does experience with LangChain, LlamaIndex, or specific LLM frameworks matter?

Familiarity with the major LLM application frameworks is useful and demonstrates that you have built things, but Arize does not require deep expertise in any specific framework. What matters more is your conceptual understanding of how LLM applications are structured, how traces and spans capture their behavior, and how evaluation can be designed to surface failures. Phoenix instruments applications regardless of framework using OpenTelemetry conventions, so the company values engineers who think at that abstraction layer rather than ones who are locked into a particular tool.

Does Arize sponsor work visas in the United States?

Arize sponsors H-1B transfers and certain other employment-based visas for specific senior or specialized roles where the candidate clearly meets a skill threshold the company cannot easily fill domestically. The company does not run a broad sponsorship program for early-career roles and does not typically sponsor candidates who would require lottery entry. If visa sponsorship is a requirement for you, raise it candidly with the recruiter during the first conversation so both sides can decide whether to invest further time in the process.

What does the take-home or technical exercise look like, and how is it evaluated?

Take-home exercises are role-specific and are scoped to a few hours of work rather than days. Engineering exercises typically involve a focused implementation task with clear acceptance criteria, while solutions-engineering exercises often involve drafting a customer-facing document or instrumentation walkthrough. Evaluators look for clean code, sensible trade-offs, clear documentation of decisions and assumptions, and evidence that you tested your work. Over-engineered submissions that ignore the stated scope are a common failure mode. Arize values engineers who solve the problem in front of them with appropriate restraint and who communicate their reasoning explicitly.

How does Arize compare to LangSmith, Datadog AI Observability, WhyLabs, Fiddler, and Galileo as a place to work?

Each of these companies occupies a different position in the AI observability and evaluation landscape, and the choice of where to work is largely about which problem space and stage you find most compelling. Arize is differentiated by its dual open-source and enterprise GTM motion through Phoenix and AX, its depth in agent and evaluation tooling, its credibility with research-forward AI teams, and a culture that pairs engineering rigor with customer obsession. LangSmith benefits from the LangChain ecosystem but couples its fortunes to that framework. Datadog AI Observability has the distribution of a public-company platform but lives inside a much larger product surface where AI is one of many priorities. WhyLabs, Fiddler, and Galileo each have specific strengths but operate at smaller scale in the enterprise market. Engineers who want to work on infrastructure that is genuinely shaping how the AI engineering discipline forms its conventions tend to find Arize the most interesting of the set.

Open Positions

Arize currently has 3 open positions.

Check Your Resume Before Applying → View 3 open positions at Arize

Related Resources

Similar Companies

Sources

Arize AI Careers (Greenhouse) —
Arize AI Official —