Careers at Beam: Teams, Pay and How to Get Hired

Who gets hired and onto which teams

Beam lists eight salaried roles across four engineering and growth functions with a $190k median band, as Zero G Talent's board data shows, and six of eight roles remote-eligible — a structure that reveals how the company organizes its GPU-native AI platform work while shipping Notes, a customer-facing product for frontline human-services workers. Beam.org says the team "changed the lives of thousands of underserved people" and is "amplifying that impact with software we build for frontline workers." Infrastructure roles exist to make that amplification possible at scale.

The AI research function centers on the Applied AI Research Engineer role, posted at $140k–$200k with remote eligibility in New York or San Francisco. This is not an applied-science title pasted onto a feature team. The role sits inside a group that trains and deploys models on the company's own GPU clusters — clusters the infrastructure team designs, provisions, and operates. Candidates who land here typically publish in top-tier venues or have shipped production models at labs where the line between research and engineering is thin. The work takes a problem from a product spec — say, summarizing case notes for a social worker, through data curation, model selection, distillation, and latency-constrained deployment. You do not hand off a notebook; you own the model in production.

The infrastructure and platform function is the largest cluster: GPU Cluster Infrastructure Engineer at $10.5k–$18k/month, Site Reliability Engineer at $140k–$200k, Distributed Systems Engineer at $140k–$200k, and Network Engineer at $130k–$180k. All four list remote eligibility. Together they form the substrate the research team depends on. The GPU cluster role focuses on scheduler topology, fabric utilization, and multi-tenant isolation — problems that look more like HPC than typical cloud DevOps. The SRE and distributed-systems roles own reliability targets for model-serving APIs that frontline workers hit in real time. The network engineer builds the interconnect that lets a training run scale across nodes without saturating the fabric. A candidate evaluating fit should ask: have I debugged a collective-communication hang at 3 a.m.? Have I designed a capacity plan for a workload whose memory footprint grows non-linearly with context length? If the answer is no, the on-ramp is steep.

Product engineering is represented by the Product Engineer role at $120k–$160k, based in New York. This team turns model outputs into the Notes interface caseworkers use. The careers page promises "high levels of ownership" and "tools you need to do your best work." In practice, you own a vertical slice: the React frontend, the API layer, the model gateway, and the feedback loop that lets a user flag a hallucination. You ship daily. You talk to users weekly. The bar is full-stack fluency plus an instinct for where model uncertainty breaks UX — and the discipline to build guardrails instead of workarounds.

Growth and operations anchors the fourth area with two roles: Head of GTM and Growth at $120k–$160k and Chief of Staff at $90k–$140k, both New York-based. The GTM role owns the motion that gets Notes into city and county agencies — a sales cycle that runs on procurement timelines, security reviews, and pilot-to-contract conversion. The chief of staff operates as a force multiplier for the founders (Luke Lombardi, Eli Mernit, Alex Frommeyer, Dan Dykes, Alex Curry), translating board-level priorities into quarterly OKRs, running hiring plans, and clearing cross-functional blockers. Both roles demand fluency in the language of public-sector buyers and the cadence of a seed-stage team that ships on Fridays.

What separates a strong application from a disqualified one is not years of experience but evidence of work that mirrors the team's actual constraints. Research wants a model you took from data to deployed endpoint. Infrastructure wants a postmortem you wrote for a GPU-cluster outage. Product wants a feature you shipped that reduced user friction measurably. Growth wants a pipeline you built in a regulated or procurement-heavy environment. The teams are small enough that every hire changes the bus factor. Candidates who treat the process as a two-way evaluation — probing the same constraints the team lives with, tend to be the ones who get through.

What it pays

These are posted bands, not negotiated outliers; they reflect what the company has committed to in public requisitions. A third-party aggregate of 139 active postings shows a $150k–$250k span and a $200k median for roles tagged "Beam" in 2026, with engineering, operations, and AI/ML leading demand.

Equity sits on top of base. The UK entity (Beam Up Ltd) advertises EMI-qualifying share options — a tax-advantaged scheme that can mean no income tax on grant and capital-gains treatment on exit. The U.S. side operates an Employee Stock Purchase Plan, letting participants buy shares at a discount through payroll deductions. Neither the board postings nor the careers pages disclose strike prices, option pool percentages, or refresh cadence. Industry practice for growth-stage companies typically issues initial grants on a four-year vest with a one-year cliff, then layers refresh grants — often called "boxcar" or "evergreen", to keep total unvested equity meaningful as the initial tranche ages. Carta says refreshes "help retain top performers and offset the costs of recruiting"; Pave and Qapita describe them as grants "designed to ensure that employees continue to have a vested interest in the company's growth and success." Ask the recruiter for the current refresh policy, the last two years' grant data for your level, and whether refreshes are automatic or performance-triggered.

Benefits fill the gap between cash and equity. The UK site lists a wellbeing budget of $400/year, enhanced parental leave at 18 weeks for primary caregivers (two weeks above the statutory minimum in some jurisdictions), a sabbatical of up to six weeks after three years, and a "work abroad" allowance of up to six weeks annually. The U.S. side (beamtx.com) cites a 5-week paid sabbatical after five years plus a $5,000 stipend, inclusive family-forming benefits of $10,000/year, financial well-being services at no cost, 401(k) with company match, tuition reimbursement, and an ESPP. A third summary — shopbeam.com, describes fully covered medical, unlimited time off, 401(k) with up to 4% match, charity matching, wellness stipend, HSA/FSA/Childcare FSA, and professional development investment. The overlap is real; the differences likely reflect entity-level plans (UK vs. U.S. subsidiaries) and page vintage. Confirm which entity holds your contract and request the summary plan description for that entity.

When you evaluate an offer, model three scenarios: base only; base plus expected annual refresh (assume zero if the company won't commit); and base plus a liquidity event at 2x, 5x, and 10x current valuation. Run the numbers on a four-year horizon with your tax rate. The median $190k–$200k base is strong for the stage; the variable is whether equity converts to liquidity on a timeline that matches your life plan. Ask for the cap table summary, the last 409A valuation date, and the board's liquidity thesis. If they can't answer, price the equity at zero and decide on cash alone.

How the hiring process works

Beam's hiring motion is lean, fast, and built around the same technical clarity that defines its product. Candidates who treat this as a volume game miss the signal: Beam publishes salary bands because it expects applicants to self-select on fit before a recruiter ever opens a file.

The first screen is a phone call. Recruiters verify baseline alignment: visa status, timeline, and whether the candidate's experience maps to the specific opening. Vague interest in "AI infrastructure" gets rejected; a concise narrative linking your distributed-systems work to Beam's GPU cluster challenges advances.

Technical evaluation follows the role. Engineering candidates should expect depth on the fundamentals that compound at scale: memory hierarchy, kernel bypass, scheduler behavior, failure-domain isolation. The board's concentration in systems roles, with five of eight openings in infrastructure or reliability, means interviewers probe for production scars, not textbook answers. A Site Reliability Engineer who can walk through a real incident postmortem, quantify the blast radius, and show how they hardened the retry logic beats a candidate who recites the CAP theorem. The GPU Cluster Infrastructure Engineer role, posted at $10,500–$18,000 monthly, signals that hardware-aware software competence is priced at a premium.

Behavioral assessment is folded into technical discussion, not a separate panel. Interviewers ask for a specific project where you hit a hard constraint, such as power envelope, latency budget, or vendor lock-in, and how you navigated trade-offs. Generic "I collaborate well" answers fail. The company's published voice emphasizes curiosity that converts to action ("I want to learn versus I think I want to learn") and self-directed project ownership. Candidates who can name the exact decision they reversed after new data, and why, signal the autonomy Beam's flat structure demands.

Speed is a feature. The process rewards reciprocity: responsive candidates who send requested references, GitHub links, or a one-page architecture sketch within hours, not days, keep momentum. Delays on the applicant side read as low intent.

Equity mechanics are standard for a seed-stage AI company: ISO grants with a four-year vest, one-year cliff. The board data doesn't publish strike prices, but the salary bands imply equity is meaningful upside, not a salary substitute. Negotiation room exists at the top of band for candidates who demonstrate unique leverage, such as open-source maintainership in a relevant subsystem, prior GPU fleet operations at scale, or a published systems paper.

Remote-first does not mean low-touch. The New York hub exists for whiteboarding sessions and hardware bring-up; distributed engineers fly in quarterly. Candidates should clarify their onsite cadence expectation before offer stage.

What disqualifies: spray-and-pray applications, inability to articulate a technical trade-off in your own words, treating the phone screen as a formality. What advances: a tailored resume that mirrors the job's keyword cluster (Engineering, AI, Platform, Distributed Systems, Agile), a 15-minute prep call with a Beam engineer via your network (recruiters hold unposted roles), and a follow-up email that references a specific technical thread from the interview. The process selects for people who prepare like they'll operate, precisely, once.

Where the work happens

The geographic split across the eight roles is the clearest signal of where the team actually works today. New York listings include Product Engineer and Chief of Staff, both tagged New York only. The remaining six openings, including GPU Cluster Infrastructure Engineer, Applied AI Research Engineer, Site Reliability Engineer, Distributed Systems Engineer, Network Engineer, and Head of GTM and Growth, all list New York, San Francisco, and Remote (US) as options. The pattern suggests a bicoastal core with remote flexibility built into engineering roles, while product and operations anchor in New York.

What the board data doesn't show is the physical nature of either site. Third-party research describes Beam Therapeutics, a Cambridge, MA biotech with a 31,000 SF fit-out at 238 Main Street and a manufacturing facility at 10 Davis Drive, Durham, NC. That is a different company. No public filing, press release, or first-party source describes the AI company Beam's office layout, lab space, colocation arrangements, or hardware access at its New York or San Francisco addresses. Candidates should treat the location tags on each posting as the only verified data point.

The role mix implies what each site enables. GPU Cluster Infrastructure, Distributed Systems, Network Engineer, and Site Reliability roles, all remote-eligible, signal that compute infrastructure and platform operations are distributed by design. Applied AI Research Engineer listings in both cities suggest research pods can form on either coast. The Chief of Staff and Product Engineer roles, New York–only, indicate that strategy, product direction, and cross-functional coordination center there. The Head of GTM and Growth role, also tri-location, reflects a go-to-market motion that can run from either hub.

Remote eligibility across six of eight roles is unusually broad for a team this size. A candidate in Austin, Seattle, or Boston can realistically target the same engineering slots as someone commuting to SoMa or Flatiron. But the bar for asynchronous communication, self-directed onboarding, and timezone-overlap discipline is higher than at companies where remote is an exception. The board data shows no roles tagged San Francisco only; every SF-eligible role is also New York–eligible and remote-eligible. That symmetry is intentional; it keeps the talent pool unified rather than bifurcated by geography.

Candidates evaluating fit should ask directly: what hardware access ships to a remote hire? What onsite days, if any, are expected for New York–anchored roles? How are research compute quotas allocated across locations? The job posts don't say. The answer lives in the conversation with the hiring team, and that conversation starts with a targeted application, not a volume play.

Who thrives here

Beam's published values, Growth, Resilience, Initiative, Tenacity, Team First, read like a filter designed for the engineering problems the company tackles. The roles on the board (GPU Cluster Infrastructure Engineer, Distributed Systems Engineer, Applied AI Research Engineer, Network Engineer, Site Reliability Engineer) all sit at the intersection of large-scale systems and bleeding-edge model deployment. That intersection selects for people who have already operated in environments where the infrastructure is the product and the failure modes are novel.

The clearest internal portrait comes from Julian Keenaghan, who joined in 2018 as employee number three and spent years as a one-person tech and product team before the organization grew past 15 engineers. Based in Ireland, he now leads technical development on Notes. His trajectory, building from zero, owning the full stack alone, then scaling the team and the architecture, mirrors the "Initiative" and "Tenacity" values explicitly. It also signals that Beam rewards engineers who can operate autonomously for extended periods without losing sight of the product outcome.

The mission language on beam.org, "build brighter welfare systems globally" and the claim of over 2,500 people supported since 2017, adds a social-impact dimension that isn't decorative. Candidates who treat that mission as a recruiting tagline tend to self-select out during the process; the ones who stay can articulate how a distributed-systems decision or a GPU-scheduling optimization connects to a downstream user outcome. That alignment shows up in the "Team First" value: infrastructure roles are not siloed; they ship in concert with product and research, and the review loops are tight.

Compensation bands reinforce the profile. The $190k median across eight salaried roles, with the Applied AI Research Engineer and Site Reliability Engineer bands reaching $200k, places Beam in the top quartile for early-stage AI infrastructure hiring. At that level, the company doesn't need to sell the role; it needs to filter for engineers who have already chosen the harder technical path. The remote-friendly designations on most listings mean the bar for communication and self-direction is higher, not lower; "Resilience" in a distributed team means you debug a cluster issue at 2 a.m. your time because the on-call rotation demands it, and you document the fix so the next engineer doesn't repeat the investigation.

Glassdoor's 3.6/5 from 123 reviews, within one standard deviation of the IT industry average, suggests the culture isn't polarizing, but it also doesn't inflate. Indeed reviews for a similarly named entity (Beam Living) are sharply negative and describe a property-management operation; they are not the same company. Beam's own employee accounts, limited as they are publicly, cluster around autonomy, technical depth, and the mission. The founders come from backgrounds that span applied research, systems engineering, and product scaling, which sets the technical tone from the top.

The four functions, research, infrastructure, product, growth, form a closed loop: research pushes model boundaries, infrastructure makes them reliable at scale, product translates them into caseworker workflows, and growth gets the result into agency procurement cycles. A candidate who can trace their own work across that loop, not just within one segment, is the one who clears the bar. The interview process doesn't test for potential in the abstract; it tests for whether you've already done the work that the next sprint demands.

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