Manifold Bio’s Million-Design Test Maps Its Six Hiring Bottlenecks

Six Open Roles, One Map of Bottlenecks

Manifold Bio has six open roles right now, clustered around three pressure points: AI platform scaling, molecular library production, and in vivo validation throughput. The roles a company funds reveal its trajectory better than any press release.

Range: The listed roles span roughly $100k to $275k. The Product Manager slot was posted four months ago, signaling an active push to commercialize the platform even as internal R&D accelerates. This hiring wave opens a rare window for candidates who can meet Manifold's exacting cross-disciplinary standards — specialists who operate at the intersection of wet-lab biology, machine learning, and high-throughput data engineering.

The highest-band role, Director of the Oligo Team, sits where synthesis scale meets data quality. Manifold's platform depends on "massively parallel molecular synthesis" — generating hundreds of thousands of barcoded protein variants per experiment. That throughput is only as good as the oligonucleotide libraries feeding it.

Two computational roles flank that synthesis engine. The AI/ML Research Engineer extended mBER, the antibody-design framework Manifold open-sourced in September 2025, which inverts AlphaFold-Multimer to optimize binders against target epitopes. The Computational Scientist, Assay Development sits closer to the wet lab, translating experimental readouts into training data. That pairing, one research-facing and one assay-facing, mirrors the "ML2" loop Manifold describes: machine learning guided by multiplexed libraries, feeding back into the next design cycle.

The wet-lab trio completes the loop. The cDNA Display scientist operates the display technology linking genotype to phenotype at library scale. The In Vivo Pharmacology study coordinator manages animal studies where Manifold's differentiator lives: testing hundreds of thousands of designs simultaneously in living systems, measuring distribution across tissues and the blood-brain barrier. A September 2025 announcement — 1.1 million VHH antibodies tested against 145 targets, over 100 million protein-protein interactions measured — is the output this role sustains. A separate March 2026 study with NVIDIA validated 1 million binder designs against 127 targets using Proteina-Complexa.

What's absent tells its own story. No clinical development roles. No regulatory affairs. No business development beyond the product-manager slot. Manifold is still building the engine, not driving the car to market. The Roche deal announced in November 2025 buys runway for that engine. The hiring plan indicates the company believes the hard problems remain upstream: better libraries, better models, better in vivo data at scale.

Inside the Engine: What Manifold Actually Does

Manifold Bio builds what it calls the first AI-guided direct-to-vivo discovery engine. Founded by CEO Gleb Kuznetsov with genetics pioneer George Church as advisor, the startup collapses the traditional bottleneck in drug development: in vivo testing. Instead of evaluating one candidate molecule per mouse — a serial process that burns time, animals, and capital — Manifold runs up to a hundred parallel tests in a single living system.

The core invention is protein barcoding. Each protein variant receives a unique molecular tag — an extra bit of protein that functions like an RFID label — making it trackable throughout a living organism. After the pooled experiment, a proprietary DNA conversion process reads out every barcode, revealing which molecules hit their target and which did not. In a typical run, equal quantities of 100 molecules enter the mouse; perhaps 95 show no activity, three bind decently, and two bind at a much higher rate. That yields 98 molecules the team can discard before spending millions on further development. Kuznetsov has compared the shift to the jump from CPU serial processing to GPU parallel processing.

The platform, dubbed mDesign, combines massively multiplexed in vivo screening with AI-powered design to create targeted biologics. "We're going end-to-end internally," Kuznetsov said in 2022. "The drugs we have in house, we created those molecules from scratch, we've put them in these pooled in vivo tests, and they'll soon be at the level of clinical trials." The initial therapeutic focus is cancer, where surface targets on tumor cells align well with the platform's strengths.

Funding arrived in two major tranches. A $40 million Series A closed in July 2022, led by Triatomic Capital with participation from Section 32, FPV Ventures, Horizons Ventures, Tencent, and existing backers Playground Global, Fifty Years, and FAST by GETTYLAB. In December 2025, an $18 million Series B led by Reach Capital, joined by SilverArc Capital, Industry Ventures, and existing investors TQ Ventures and Calibrate Ventures, explicitly targeted "agentic AI capabilities" — what the company calls Agent OS — a layer of agents operating on life sciences data types including molecular, phenotypic, clinical, and real-world data.

Market validation came via a strategic collaboration with Roche, reported as a potential $2 billion pact. The agreement grants Roche access to Manifold's tissue-targeting shuttle portfolio and the mDesign discovery engine, signaling that a top-tier pharma player views the platform as a viable route to brain and other tissue targets.

Manifold operates from the Boston area with roughly 60 people. The Series B capital and Roche partnership suggest the next hiring wave will lean heavily into the Agent OS stack, roles at the intersection of large-scale biological data and autonomous AI workflows. That demand shapes the screen candidates face today.

The Screen: What Clears the Bar

Manifold's screen filters for candidates matching that hybrid profile. The six roles map directly to the three pillars of its platform: AI-guided protein design, massively multiplexed in vivo screening, and the data infrastructure binding them.

The AI/ML Research Engineer and Computational Scientist, Assay Development positions demand fluency in model training, experiment design, and the peculiar noise profiles of biological data. The cDNA Display and In Vivo Pharmacology roles require hands-on expertise with library construction, animal studies, and the logistical complexity of running 100-compound screens in parallel. The Product Manager, AI Platform & Partnerships sits at the translation layer, someone who can speak to external partners and internal engineers without losing fidelity in either direction. The Director, Oligo Team Lead owns the oligonucleotide supply chain feeding the entire engine. None of these roles tolerate siloed expertise.

Publicly documented interview loops from InterviewQuery's Data Engineer candidate reports show technical, analytical, and communication-focused questions covering data modeling, ETL pipeline design, cloud infrastructure, and presenting complex scientific data to diverse audiences. That last item, communication across disciplines, appears repeatedly in Manifold's own messaging. Kuznetsov, a former Church lab graduate student, has described the platform as an "in vivo design engine" generating "unprecedented data on the in vivo targeting behavior of our drug candidates."

Beyond role-specific requirements, the screening environment has shifted. A 2025 industry analysis notes that HR teams now use generative AI tools to filter resumes by keyword correlation (skills, experience, location) before a human sees the document. The same analysis estimates 60% of biotech HR departments still rely on Excel-based manual review, but the trend accelerates toward automated shortlisting. Candidates who treat their CV as a narrative document rather than a structured keyword index (technical skills tabled with years of experience, project relevance highlighted, awards and publications surfaced) lose the algorithmic round. Cover letters function as a distinct evaluation layer: hiring managers read them for communication clarity and domain specificity, and they detect fully AI-generated text.

Manifold has not published a competency framework, scoring matrix, or diversity slate requirement. Job descriptions list qualifications in broad strokes ("PhD or equivalent experience," "proficiency in Python and ML frameworks," "experience with in vivo studies") without weighting or knockout criteria. Interview process length, panel composition, and take-home assignment expectations are not public. Candidates should prepare for a process testing both depth in their primary discipline and literacy in the other two pillars of the platform.

Applying: Mechanics and Pitfalls

Manifold lists open roles on its careers portal, LinkedIn, and specialist boards including Zero G Talent. Four of six roles are Boston-only; the two hybrid roles (Product Manager and AI/ML Research Engineer) explicitly list Boston or San Francisco. Remote-only applicants will not clear the first pass for Boston-locked positions. Salary bands are wide, up to $110k spread on the Product Manager role, suggesting leveling is negotiated per candidate rather than fixed to a grade. Come prepared with a concrete number backed by comparable frontier-biotech offers.

The job descriptions themselves are the clearest signal of what the screen seeks. The Product Manager posting asks for someone who can "define the product vision, roadmap, and success metrics for Manifold's AI platform capabilities, including protein design, foundation models, and agents" and "own Manifold's strategy for engaging the AI ecosystem; decide which partners to prioritize, what to build with each, and how to sequence engagements." Wet-lab roles (cDNA Display, Assay Development, In Vivo Pharmacology) all sit in Boston and call for hands-on experience with the exact workflows Manifold runs: high-throughput molecular barcoding, pooled in vivo screening, and the downstream data pipelines turning those reads into design loops. The AI/ML Research Engineer role spans both coasts and expects fluency in the same stack: foundation models for protein design, agents for experimental planning, and the compute infrastructure to run them at million-experiment scale.

Manifold has not published a public interview rubric. In the absence of company-specific guidance, candidates should expect a standard deep-tech biotech loop: initial recruiter screen, a technical deep-dive with the hiring manager or a senior scientist/engineer, a cross-functional panel (wet-lab leads, computational leads, and often a product or strategy voice), and a final conversation with a co-founder or VP. The company's public emphasis on "cross-disciplinary standards" means you will likely be evaluated by people outside your primary discipline: a computational candidate should expect wet-lab questions about assay noise and in vivo variability; a bench scientist should expect to discuss how their data feeds model training.

Two practical pitfalls appear in the specs. First, several roles list "PhD or equivalent experience" but the equivalent bar is set by specific techniques: mCodes barcoding, cDNA display libraries, pooled in vivo pharmacology, and large-scale protein-language-model fine-tuning. Generic "AI for biology" publications will not substitute. Second, the Product Manager role demands ecosystem-partnership judgment: selecting priority partners, defining co-build scope, and sequencing engagements. That is a business-development motion disguised as product management; candidates who have only shipped internal tools at large pharma or pure-software shops will struggle to demonstrate the required context.

No public source documents a referral program, take-home assignment, or coding challenge format. The safest path: apply directly through the Manifold careers page or the Zero G Talent listing (which links to the same ATS), tailor the resume to the exact technique keywords in the posting, and be ready to walk through a specific project where you closed the loop between molecular design, in vivo data, and the next design iteration. That loop is the product Manifold sells; showing you have already run it is the only credential the screen cannot ignore. The broader market context explains why that credential is scarce.

What This Hiring Wave Signals for Biotech

Manifold's six open roles read like a taxonomy of the 2026 biotech labor market: computational, wet-lab, translational, and product, exactly where the industry's scarce hybrid talent sits.

The market isn't frozen. It's selective. BioSpace's 2026 US Life Sciences Employment Outlook found 64% of surveyed biopharma organizations actively recruiting and 41% expecting more open roles this year. Yet CBRE reported lab and R&D vacancy at 23.2% across the top 13 US life sciences clusters in Q1 2026, even as biotech R&D employment rose for a fifth straight month to a record level in February. Broader life sciences employment fell 0.4% year over year. The contradiction resolves when you look at what is actually being filled: roles that move programs toward IND, BLA, and commercial supply (regulatory, CMC, clinical development, manufacturing operations, and the computational biology feeding them). Manifold's slate sits in that band.

The layoff backdrop sharpens the picture. BioSpace counted roughly 42,700 US biopharma professionals affected in 2025, up from about 29,000 in 2024. VectorTA's April survey found over half of currently employed candidates actively looking and 85% of unemployed ones weighing roles outside biopharma entirely, with one respondent stating there is zero job security in biopharma. Candidates who have lived through cuts (or watched colleagues go) now scrutinize runway, pipeline stability, and leadership before they engage.

Manifold enters this environment with a fresh Series B and a platform multiplexing protein measurement in vivo. That stability signal matters. VectorTA notes that hiring managers need to be more forthcoming about financial position and what genuine job security looks like. Companies that engage good people early and are honest about runway win the hires that matter.

Cell and gene therapy attracted $15.2 billion in investment in 2025, up 30% on 2023, but the talent infrastructure to support it has not grown at the same pace. Bioinformatics and computational biology face the same squeeze: roles increasingly demand a hybrid of deep domain science and data or programming skills that relatively few people have built over a career. Manifold's Computational Scientist, Assay Development role (requiring both assay biology and code) sits in that exact gap. So does the AI/ML Research Engineer who must understand the wet-lab constraints of the platform.

Contract and fractional hiring has become structural. GForce Life Sciences reports 30 to 40% or more of critical work delivered through contractors, even at organizations that once defaulted to permanent hires. LinkedIn's 2026 trends note the same: many organizations lean on contractors and fractional leaders to preserve flexibility, ramping up around milestones without locking in long-term headcount. Manifold's six salaried roles signal a commitment to core team building rather than pure project-based staffing.

Location remains a hard filter. Biotech is more location-dependent than most industries. The share of employers willing to recruit regardless of location dropped from nearly 50% in 2022 to around 20% in 2024. Most now prefer local candidates with hybrid arrangements for roles not requiring daily lab access. Manifold lists Boston or San Francisco for product and AI/ML roles, Boston for wet-lab and computational roles, aligning with the two hubs still leading hiring volume, per LinkedIn, while San Diego, Research Triangle Park, Boulder and Denver, and parts of Texas and North Carolina see sustained growth.

BLS projections through 2034 reinforce the directional bet: data scientist roles up 34%, computer and information research scientists up 20%, operations research analysts up 21%, medical scientists up 9%, epidemiologists up 16%, industrial engineers up 11%. The computational-biology hybrid is not a fad. It is the growth vector.

For candidates, the practical implication is clear: target areas where hiring is genuinely happening (regulatory, clinical, manufacturing, and the computational-experimental interface) rather than searching broadly. Emphasize concrete, in-demand skills that move programs forward. Network relentlessly, since much hiring happens through connections and being visible when companies quietly rebuild matters more than waiting for job boards to fill. Scrutinize prospective employers' financial health and pipeline, because candidates today are right to weigh stability, not just the role.

For companies, the lesson is that the talent bench is thinner than it looks. Organizations building the best teams over the next 12 to 18 months will treat recruitment as a long-term function rather than a reactive one: mapping future talent needs against pipeline milestones, building relationships before vacancies open, and being honest with candidates about what they are joining. Manifold's current search (platform engineering, assay development, in vivo pharmacology) looks like a company doing exactly that. The loop its new hires will close is the same loop the platform runs.

Boundaries: What This Article Does Not Cover

This piece defines its boundaries the way a project manager defines a work package: by naming what sits inside and, just as deliberately, what sits outside. The exclusion list is not a deficiency. It is the boundary that lets the included material hold its weight: the same loop Manifold runs, now closed on the page.

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