Slingshot AI Pays Median $307K for AI‑Psychology Hybrid Roles

Five Roles, One Bet

Slingshot AI listed five open positions in August 2026: a postdoctoral clinical researcher, a technical ex-founder, a mobile engineering lead, a chief of staff for go-to-market, and a science-communication intern, all targeting the same intersection: engineers who speak therapy and clinicians who code. The New York– and London–based startup behind the Ash therapy chatbot has raised $93 million total, including a $53 million Series A in July 2025 led by Radical Ventures and Forerunner Ventures, and operates with 11 to 50 people. Slingshot AI lists four salaried roles spanning $86,000 to $338,000 (median $307,000) plus one internship.

The most senior technical opening, Member of Technical Staff, Mobile, will own the iOS and Android surfaces that deliver Ash to consumers — a direct-to-consumer play the founders have said they intend to keep "ideally forever" because it forces product quality. The Technical Ex‑Founder role, same office and band, targets founders who have shipped complex products and want to apply that zero-to-one experience to a mental health foundation model rather than chase AGI.

In New York, the Chief of Staff, GTM mandate is broad: work across strategy, operations, and partnerships as the company prepares to scale a consumer subscription product the founders have compared to Netflix or Spotify pricing, roughly $10 to $20 per month. The Postdoctoral Clinical Research Scientist role carries a $65,000–$80,000 range. This is not a token science hire. The lab says it has built the largest dataset of its kind, with over 50,000 users opting in to share data for model training. The postdoc will design and run studies that validate Ash's clinical efficacy — work that distinguishes Slingshot from general-purpose chatbots like ChatGPT or Claude, which the company explicitly contrasts as "assistants" that validate rather than challenge users.

Rounding out the five is the Social & Creator Intern, focused on translating the science behind Ash for a consumer audience, an unusual priority for an early-stage AI lab, but consistent with a direct-to-consumer strategy where trust and comprehension drive adoption.

Together, the five openings form a coherent picture: clinical research to ground the model, mobile engineering to deliver it, a founder-grade technical lead to architect it, a go-to-market operator to distribute it, and a science communicator to explain it. No single role carries the full weight. The hiring sprint signals a bet that the hard problem in mental health AI isn't just model architecture — it's the integration of clinical validity, consumer product quality, and distribution into one team.

What the Screening Process Values

Slingshot's hiring signals treat machine-learning depth and clinical fluency as co-requirements, not complementary nice-to-haves. Public statements, team composition, and open roles form a consistent picture: candidates who cannot operate at the intersection of large-language-model training and therapeutic practice face a steep climb.

Technical bar: production-grade LLM experience

The technical criteria are explicit in how the team describes its own workflow. In a November 2025 interview, the founders outlined a three-stage pipeline — pre-training, alignment via direct preference optimization, and reinforcement learning, supported by rapid synthetic-data generation and a "behaviors" taxonomy for tagging model outputs. That stack implies the screening process filters for engineers who have shipped DPO or RLHF loops, built evaluation suites for generative models, and worked with synthetic data at scale. Both London technical roles carry £100K–£250K bands with equity, compensation that aligns with senior IC or founding-engineer expectations rather than junior hiring. First-party board data shows a median salary band of $307K across four salaried roles, reinforcing the seniority signal.

Glassdoor lists three interview questions and three reviews, a small but telling sample suggesting a selective, low-throughput process rather than high-volume screening. The company's own careers page notes: "We tend to be quite a senior team. Most folks have been managers in a past role." That self-description functions as a de facto filter — candidates without lead or ownership experience are unlikely to clear the initial review.

Domain bar: clinical literacy, not just curiosity

Slingshot's model, Ash, is trained across CBT, DBT, ACT, IFS, psychodynamic therapy, motivational interviewing, and somatic therapy. The company maintains a full-time clinical team that produces DPO examples, preferring one model response over another, and tags unacceptable behaviors to build evals. That workflow demands candidates who can speak the language of therapeutic alliance, transference, and modality-specific technique without hand-holding.

That role makes the requirement concrete: a PhD or PsyD with research background, comfort with IRB processes, and the ability to translate clinical judgment into model-alignment signals. Chief of Staff, GTM sits at the commercial-clinical interface, requiring someone who can articulate the product's therapeutic rationale to payors, providers, and regulators. Even the Social & Creator Intern role emphasizes science communication, translating peer-reviewed concepts for a lay audience without distortion.

Values as behavioral filters

Five published values operate as behavioral screens. "Always in search of truth" maps to a culture of open debate and evidence-following — candidates who optimize for consensus over correctness will struggle. "Move fast and embrace the unknown" selects for engineers comfortable shipping imperfect models into a high-stakes domain where a hallucinated safety response carries real risk. "Celebrate divergence and value authenticity" signals tolerance for non-traditional backgrounds, but only when paired with demonstrated rigor. "On a mission, and the clock is ticking" filters for urgency; about 169 million Americans live in mental-health workforce shortage areas, per HRSA data from December 2023, the backdrop against which every hire is measured.

The dual-skill synthesis

The screening process operates as a synthesis test: can the candidate take a clinical concept — say, the therapeutic rupture-and-repair cycle, and design a DPO dataset that teaches the model to recognize and respond to it without defaulting to sycophancy? Can they debug a "behavior" tag that flags premature advice-giving in a motivational-interviewing context? The roles, the team's self-description, and the technical architecture they've made public optimize for that synthesis. Applicants who lead with only one side of the equation, deep ML without clinical exposure, or clinical credentials without hands-on model work, are competing against a bar set for the intersection.

How Candidates Are Tailoring Applications

Machine-learning roles have normalized a playbook candidates now run: mirror the job description's keywords, quantify every project outcome, strip formatting quirks that trip applicant-tracking systems. Enhancv's 2026 guide to ML resumes stresses that most employers use ATS to filter CVs and advises applicants to incorporate relevant keywords while using numbers to detail contributions. Slingshot Resume, an AI-powered builder, echoes that tailoring ensures key skills and terms are included and that AI helps resumes meet those criteria, increasing visibility. This baseline discipline is now standard for any technical role.

Slingshot's open roles, however, demand a second layer of tailoring that generic ML advice doesn't cover. The clinical research role explicitly signals a need for domain fluency: "postdoctoral" and "clinical" are not keywords a pure ML engineer typically leads with. The technical-ex-founder and mobile-engineering slots index heavily on shipping product at speed. Candidates who only speak one language, either therapy protocols or transformer architectures, are learning to lead with the hybrid.

Evidence of that repositioning appears in how applicants structure project narratives. Enhancv's standard ML resume template recommends candidates focus on specific accomplishments to show real-world impact and list relevant certifications. For Slingshot aspirants, "real-world impact" increasingly means a pilot study with a community mental-health clinic, a fine-tuning run on de-identified therapy transcripts, or a collaboration with a university counseling center — artifacts that sit at the intersection of the two skill sets. Certifications have shifted too: alongside AWS ML Specialty or TensorFlow Developer certificates, candidates now surface clinical-research credentials (CITI training, IRB experience) or peer-reviewed publications in journals like JMIR Mental Health or Internet Interventions.

The ATS constraint hasn't disappeared; it has compounded. Slingshot Resume notes the same ATS screening practice and the AI's role in meeting criteria. But the keyword set has expanded. Where a standard ML posting might surface "PyTorch," "distributed training," and "MLOps," Slingshot's descriptions, inferred from role titles and the stated mission around the Ash foundation model for psychology, likely also weight "clinical validation," "therapeutic alliance," "safety monitoring," and "HIPAA-compliant data pipelines." Candidates who stuff only the former keyword cluster risk passing the first filter but failing the human review that follows.

Career-switchers report a steeper translation burden than in prior ML hiring waves. Enhancv's guide flags that newcomers often struggle with format, highlighting limited experience, and using impactful language. For a clinician moving into ML, "limited experience" means no production model deployments; for an engineer moving into mental health, it means no IRB protocols or supervised practicum hours. Strong applications frame the gap as deliberate cross-training: a clinician who completed a 13-month ML engineering program (UCR's online track advertises expedited completion in as little as 13 months) or an engineer who volunteered on a crisis-text line while building a side project that classifies suicide-risk signals.

What the research does not yet capture is a verified sample of successful Slingshot applications: no anonymized resumes, no hiring-manager debriefs, no applicant surveys specific to this cycle. The board data confirms the role mix and compensation bands; public resume guides confirm the mechanical tactics candidates are drilled on. The dual-skill signaling, clinical credibility paired with engineering rigor, is the logical inference from both, but direct evidence of applicant behavior at Slingshot remains anecdotal. As the hiring sprint progresses, the clearest signal will come from who gets interviewed: whether shortlists tilt toward clinicians who code, engineers who counsel, or a new hybrid profile that didn't exist in the talent pool two years ago.

Regulatory Pressure Reshapes the Talent Market

Regulatory pressure across state capitols is reshaping what mental-health AI companies need to hire for — and what candidates must prove they can do. As of June 2026, 78 chatbot-related bills have been introduced in 27 states, per the Transparency Coalition's legislative tracker. Vermont Governor Phil Scott has already signed a therapy chatbot ban into law. Missouri's SB 1019 prohibits offering AI therapy chatbots and carries a $10,000 fine for a first offense, $20,000 for subsequent violations. New Hampshire's SB 640 would bar AI systems from posing as state-licensed counselors. Pennsylvania, Rhode Island, Hawaii, and California have parallel measures moving through committee. The legislation creates immediate demand for engineers who can build guardrails, clinicians who can validate safety protocols, and product leads who can manage compliance before a model ships.

Slingshot's own board data reflects that pressure. The open positions span product, engineering, and clinical research. The spread signals a market that prices dual fluency at a premium. A pure ML researcher without clinical exposure sits at the lower end; a founder-level hire who can steer both model architecture and regulatory strategy commands the top.

The legislative tracker shows bills like Hawaii's SB 3001 requiring suicidal-ideation prevention protocols and minor-user protections, and Pennsylvania's HB 2006 mandating safeguards for companion applications. Each requirement translates into a hiring criterion, safety-engineering experience, IRB-familiar clinical staff, regulatory-affairs fluency, that narrows the candidate pool.

Research on downstream hiring trends is thin. No public dataset tracks how many mental-health AI roles have been posted, filled, or re-scoped since the bill surge began. But without systematic data, the magnitude of the shift remains qualitative. What is clear: the regulatory environment has turned domain knowledge from a nice-to-have into a compliance necessity, and the salary bands at Slingshot and the breadth of its open roles are an early market signal that the talent market is repricing accordingly.

Where Regulation, Science, and Clinics Collide

The FDA's Digital Health Advisory Committee convened in November 2025 to wrestle with a question that has no precedent: how to regulate therapy chatbots powered by generative AI. The meeting was explicit about the stakes — large language models produce conversation-like outputs that are not predictable and may misguide users or lead to patient harm. Committee members were asked to consider the device for both prescription and over-the-counter contexts, for adults as well as adolescents, and with indications for major depressive disorder and multiple mental health conditions. The feedback will inform the agency's broader work on AI regulation, but no framework exists yet.

The American Psychological Association has not drafted guidelines for AI therapists. APA's Wright told Healthcare Brew the organization plans to start soon. In the meantime, APA is urging federal regulators to put out guidelines and require apps to have safety measures like crisis support. The group advocates for further research and says AI should be a tool, not a replacement, for human care. A 2018 meta-analysis found the alliance-outcome link explains roughly 7% of variance with an average effect size of .26, robust across multiple meta-analyses.

Academia is moving faster than policy. Brown University announced in July 2025 it would lead a research institute focused on AI assistance in behavioral health, backed by $20 million from the National Science Foundation. The institute aims to study how AI can support, not supplant, clinical work. Dr. Mark Ungless, a psychologist and director of data science, AI, and research at Mental Health Innovations, advises on clinical validation design. Dr. Thomas Insel, who ran the National Institute of Mental Health from 2002 to 2015, shapes regulatory strategy. Patricia Arean, former research director at NIMH, oversees trial design. Former Congressman Patrick Kennedy and Dr. Devika Bhushan, former Surgeon General of California, provide policy counsel. Mahmoud Khedr brings lived-experience advocacy. Their presence signals that Slingshot is courting institutional credibility, not just venture validation.

Providers remain cautious. Healthcare Brew reported tales of AI stoking delusions and chatbots telling a user "to get rid of" his family. Regulators are struggling to keep up as use grows. About half of psychologists had no openings for new patients in 2024, and more than half of people struggling with mental health receive no care at all, pressure that no advisory board can resolve alone. Slingshot's board members have publicly framed Ash as expanding access to evidence-based tools, not replicating traditional therapy in AI form. But the gap between "expanding access" and "meeting clinical standards" is where the next regulatory fight will land.

The five roles Slingshot posted in August, mobile engineer, ex-founder, chief of staff, postdoc, intern, each hold a piece of the puzzle: a model that must be clinically rigorous, technically novel, and consumer-ready all at once. The postdoc designs studies on 50,000 opted-in users. The mobile engineer ships the interface that delivers Ash to a phone screen. The ex-founder architects the DPO loop that teaches the model therapeutic rupture and repair. The chief of staff manages the $10-to-$20 subscription past state regulators. The intern translates the science so a user in a shortage area understands what they're trusting. The next regulatory fight lands in that gap. The talent market has already priced the answer.

Working in frontier tech? Zero G Talent tracks the openings: see every open Slingshot Aerospace role, browse frontier tech jobs, openings at Slingshot AI, and the people building the field.