How Work Actually Gets Done
A Cambridge biotech says it designed, built, and tested five antibody programs end-to-end in roughly six weeks in one parallelized campaign, with hit rates in the double digits rather than the fractions of a percent that traditional discovery methods post. That figure from Nabla Bio's own LinkedIn account captures more about the company's claimed pace than any org chart could. The work Nabla describes doesn't unfold in tidy quarterly milestones; it moves in compressed cycles where design, wet-lab testing, and model retraining run in the same sprint.
The structure that enables this, per Nabla's own platform page and careers text, is intentionally flat. Built In Boston lists Nabla Bio as having 11–50 employees; LinkedIn lists 10,953 followers and 26 employees visible on the page; and Tracxn reports 23 employees as of June 2026, with a Cambridge, Massachusetts headquarters. Since launching in 2021 the company has raised $37 million from Radical Ventures, Khosla Ventures, and Zetta Venture Partners. With that headcount, decision-making authority sits close to the bench. Surge Biswas, the co-founder and CEO, sets the technical direction in public remarks, telling Genetic Engineering & Biotechnology News in February 2026 that "this technology is ready for prime time and should be a first-line approach in drug discovery campaigns," but the operating rhythm described across Nabla's platform and LinkedIn posts shows authority handed to whoever owns the assay, the model output, or the campaign result.
The cadence reflects what Nabla calls its integrated AI-and-wet-lab loop. JAM, the molecular design model, generates antibody candidates in silico, and Nabla's mammalian expression systems measure binding, developability, and function in cellular and in vivo contexts that mirror human biology. Those measurements feed back into the model. The platform page describes this as a single tightly integrated data flow designed to "sharpen the model's ability to design molecules with human-ready properties from the start, eliminating the need for iterative lab-in-the-loop screening." In practice, per the company's own framing, an ML researcher and a bench scientist can co-own a target by Tuesday and ship candidates to a pharma partner's discovery team by Friday.
The pharma partnerships shape the workflow in concrete ways. Nabla's company description lists research collaborations with AstraZeneca, Bristol Myers Squibb, and Takeda. The Takeda deal, announced in November 2025, grants Nabla eligibility for success-based payments reaching upwards of $1 billion and spans de novo antibodies, multispecifics, and other custom therapeutics. At the J.P. Morgan Healthcare Conference in January 2026, Takeda president Andy Plump described a Nabla-designed molecule that rescued a program his team had been about to terminate: "Could we have come up with this ourselves using traditional approaches? Never in a million years." A project of that sensitivity, Nabla's materials imply, can't tolerate handoffs across five departments.
The salary structure on Zero G Talent's Nabla Bio listings confirms the shape of the org in dollar terms. Across the board's Nabla Bio postings, the median sits near $180,000, with bands stretching from a Mid-Market Account Executive floor of $90,000 to a senior backend ceiling of $220,000. There's no separate "associate scientist" rung visible in the board data, which is consistent with the all-hands-on-deck pace the company describes: the modelers and the bench scientists are framed as the same bilingual group.
Values and Operating Principles
Nabla Bio compresses its operating philosophy into one sentence on its team page, and that single sentence does most of the work that a values poster does at larger companies. The company says its culture "prizes speed, creativity, and rigor," then unpacks each of those words with a specific behavioral rule: debate ideas with data, not ego; value steep growth over static titles; and work in person, shoulder-to-shoulder in the lab every day. Read together, those three clauses form a coherent operating system. The company is telling candidates, in effect, what the daily norm is going to feel like and what deviation will be noticed. Each principle doubles as a screening filter, which is why the hiring bar reads the same way the values page does.
The first principle, debate ideas with data, not ego, is the load-bearing one. When the customer list includes three top-20 pharmas and the science depends on whether a de novo antibody actually binds the intended epitope, the cost of letting seniority or charm win an argument is unusually high. The "data, not ego" rule exists because the model and the wet lab have to agree before a candidate moves forward. The careers page frames the implication plainly: "We care about getting to the truth, building technology that works, and learning quickly from the results." A scientist who can't cite their evidence, or can't update when the evidence updates, slows everyone down.
Speed and rigor are paired deliberately, because in antibody discovery those two values usually trade off. Nabla's pitch is that they don't have to. Its JAM platform is built to generate drug-like candidates straight from computational design (the announcement of JAM-2 appeared on Nabla's site, with a technical report linked from the homepage), and the company frames its human-relevant assays as the rigor layer that lets the speed be defensible. Genetic Engineering & Biotechnology News reported in early 2026 that JAM-2 achieved double-digit hit rates from small design sets across 28 structurally and functionally diverse targets, including G protein–coupled receptors, a significant advance over traditional methods where success rates land at less than one percent. Endpoints News reported in November 2024 that Nabla was producing "de novo antibodies [that] look more like drugs," meaning candidates with selectivity and developability characteristics good enough to interest partners — not just computationally plausible binders. The operating principle, translated into Monday-morning behavior, is: ship a candidate fast, then kill it fast if the assay says no.
The steep growth over static titles clause is the one most candidates underestimate on the way in. The team page says employees should expect to expand their scope faster than their job description. At a Series A company with revenue and Big Pharma collaborations already on the books, the person who joined as a bench scientist in 2024 may own a partner-facing readout by 2026. The flip side is that there is no large org chart to hide inside; titles are descriptive, not protective.
Finally, the in-person, shoulder-to-shoulder lab requirement rules out remote work entirely. Built In Boston notes that Nabla Bio employees work from physical offices ("OnSite Workspace"), with the typical time on-site at the Cambridge headquarters, and the team page ties the cultural norm to physical co-location. For wet-lab work this is unsurprising; the more pointed implication is that dry-lab ML people are also expected at the bench. The integration of AI and wet-lab teams, which Nabla describes as a single bilingual group of scientists, is hard to maintain over Slack.
What the Hiring Bar Selects For
Nabla Bio's hiring funnel is short on ceremony and long on evidence of doing. The careers page doesn't list a formal interview rubric; candidates email [email protected] with a CV and a short statement of interest. But the signals that move an application forward track closely with how the company says it works: scientific depth, bias toward action, and a willingness to operate in ambiguity.
Glassdoor data pegs the process at roughly 41 days on average across 6 submitted interviews, with candidates rating overall interview difficulty 3 out of 5. That's a mid-length, mid-difficulty bar, meaningful, but not the multi-round gauntlet common at larger pharma partners. In a startup that says it can generate a million antibody variants and read out their target interactions in two to three weeks, hiring managers can't afford a six-month search.
What they're selecting for falls into three overlapping buckets.
First, credible technical chops at the intersection of machine learning and biology. Nabla is a spin-out from George Church's lab at Harvard, and its cofounders (Surge Biswas, from Church's protein-engineering group, and Frances Anastassacos, from William Shih's DNA-origami lab) built the company around that hybrid pedigree. JAM, the platform behind their pharma collaborations, was trained on what Nabla describes as "massive protein sequence and structure data" strengthened with Nabla's own human-relevant measurements. Candidates who can show real work in protein engineering, generative models for biology, or structure-based design have a structural advantage over generalist ML applicants. Job postings currently live on the Zero G Talent board for an SRE / Backend Engineer at $160,000–$220,000 and a Technical Product Manager — EHR at $140,000–$180,000. Both of these sit on the engineering side of the platform.
Second, scientific judgment under uncertainty. The company's public posture is explicit about this: The phrase "learning quickly from the results" matters more than it reads. Nabla publishes preprints before peer review, including a 2025 BioRxiv paper on dozens of antibody candidates designed against G protein–coupled receptors. Interviewers screen for people who can run a wet-lab or computational experiment, accept a null, and reroute without losing a week. A candidate prep video on YouTube (a general interview-advice channel, not a Nabla-specific source) summarizes the underlying signal bluntly: avoid "I enjoy science" or "I enjoy working in a lab." The interviewer wants to hear what sets you apart from the next applicant.
Third, integrity that holds when speed tempts shortcuts. The same general interview-advice video flags ethical decision-making as a hiring signal because "a poor ethical decision can result in the invalidation of testing and lost time and money for the company." For a platform whose antibody designs are optimized for manufacturability and developability before they ever leave the computer, data integrity isn't a soft skill — it's the difference between a candidate that reaches patients and one that costs a partner like Takeda months of timeline.
Not every signal the bar sends is consistent. One Glassdoor reviewer praised HR ("friendly, responsive, and gave detailed, constructive feedback after the interview") but added that technical interviewers could use better training and called for the session to be more open-ended. Candidates who thrive are the ones who treat that ambiguity as the test, not a bug; the same posture the work itself demands once they're in.
What Current and Former Employees Say
Public employee feedback on Nabla Bio is thin, and that thinness is itself the most honest thing you can tell a candidate. As of mid-2026, the company's primary Glassdoor profile carries a 4.0 out of 5 rating across 17 anonymous reviews. On Glassdoor's interview section, 28 interview reviews sit alongside 26 interview questions, a ratio that signals something specific: more people have documented the hiring process than the day-to-day experience, which means the employee-side record skews toward people evaluating the company rather than people staying in it.
What does exist in those 17 reviews leans positive on a few recurring axes. Reviewers consistently cite the technical caliber of colleagues and the pace of experimental iteration as the strongest parts of the role, language that tracks closely with the operating principles the company publishes about rapid wet-lab cycles and high ownership. Several describe leadership as accessible and decisions as being made close to the bench, a comment that lands differently depending on which lab or function the reviewer sits in. Compensation, where it appears, is described as competitive but not outlier, which aligns with the salary band visible on Nabla Bio's current postings.
The criticism in that same Glassdoor set is narrower but worth naming precisely. The most frequent negative thread concerns workload and the cost of Nabla's stated bias for speed: reviewers describe weeks where priorities shift mid-stream and where the line between "rapid iteration" and "relentless" blurs depending on how the reviewer reads the operating principles. A smaller cluster flags ambiguity around long-term career paths, typical of a company still scaling its first commercial platform, and harder to fix with process alone. One or two reviews mention communication gaps between the wet-lab and computational teams, a friction point that the company publicly frames as a feature ("model designers stay close to experimentalists") but that some reviewers experience as the same principle applied unevenly.
LinkedIn and other first-person accounts exist but are harder to verify at the individual level. Current and former Nabla Bio employees post about the company in their headlines and feeds, and the public commentary there tends to repeat the Glassdoor themes rather than contradict them. If you are weighing the company on its own testimony, treat the Glassdoor 17-review set as the densest evidence you have, read the negative reviews in full (they are short), and weight the most recent ones more heavily; small samples shift fast.
Who Thrives Here and Who Burns Out
The traits that map to long-term fit at Nabla Bio stack up predictably once you read the public material alongside the job board. The company operates inside a roughly 23-person organization (Tracxn, June 2026) running five parallel antibody campaigns on a roughly six-week cycle, with pharma partners like Takeda eligible for success-based payments north of $1 billion. That kind of compression rewards people who default to action over consensus and who can absorb shifting priorities without losing a week.
| Role | Band (USD/year) | What the band signals |
|---|---|---|
| SRE / Backend Engineer | $160,000–$220,000 | Senior, scarce skill; ceiling reserved for proven operators |
| Technical Product Manager – EHR | $140,000–$180,000 | Domain-loaded PM seat, not a generalist slot |
| Growth Marketing Director | $140,000–$180,000 | Director-level scope without principal-tier premium |
| Partnership Director | $120,000–$160,000 | Floor above AE; expects self-sourced pipeline |
| Mid-Market Account Executive | $90,000–$120,000 | Tight band — paying for sellers, not raw potential |
The burnout profile is easier to read because Nabla Bio's own staffing data keeps writing it down. The same Glassdoor 17-review set that praises the technical caliber and pace of iteration also surfaces the strain: reviewers describe weeks where priorities shift mid-stream, and the line between "rapid iteration" and "relentless" blurs depending on how the reviewer reads the operating principles. Candidates who prize deep specialization over breadth, who want a nine-month roadmap rather than a six-week loop, or who need visible hierarchy to know what to do next will feel that friction before they hit their first anniversary. The smaller cluster flagging communication gaps between wet-lab and computational teams, the very bilingual integration the company markets as a feature, points to where the in-person, shoulder-to-shoulder requirement starts to chafe.
The honest tension to flag: nearly all the public signal about Nabla Bio's culture comes from the company's own press releases, its partners' testimonials, and the Glassdoor 17-review set. There is no review volume large enough to triangulate attrition rates, promotion timelines, or average tenure. The prudent read is that anyone evaluating fit should weight the operating principles, the partner deal terms, and the salary bands more heavily than employee sentiment, because the employee-sentiment sample simply isn't thick enough yet to support a confident read. Six weeks, five programs, three pharmas — that's the loop the values page is asking new hires to run on repeat, and the same loop that decides, well before a first performance review, who stays.
Working in frontier tech? Zero G Talent tracks the openings: see every open Nabla Bio role, browse frontier tech jobs, the companies hiring, and the people building the field.