Helium's Current Talent‑Dense Hiring Push
Helium, an eight‑person AI startup serving tens of millions of users daily with revenue doubling month over month, has opened six roles at once — a sprint that would strain a company twice its size. The roles carry "Founding" tags or sit directly beside the founders: Senior Full‑Stack Engineer ($150K–$250K, 0.50–1.50% equity), Full‑Stack Engineer ($125K–$200K, 0.50–1.00%), Design Engineer ($150K–$200K, 0.50–1.00%), Account Executive ($140K–$180K, 0.15–0.30%), Chief of Staff ($120K–$160K, 0.30–0.60%), and Growth Operations ($80K–$150K, 0.20–0.40%). Shray Arora is the hiring contact for each role listed on the company's hiring portal; the careers page shows all six roles live with apply links to Y Combinator's job board.
| Role | Salary | Equity |
|---|---|---|
| Founding Senior Full‑Stack Engineer | $150K–$250K | 0.50–1.50% |
| Founding Full‑Stack Engineer | $125K–$200K | 0.50–1.00% |
| Founding Design Engineer | $150K–$200K | 0.50–1.00% |
| Founding Account Executive | $140K–$180K | 0.15–0.30% |
| Chief of Staff | $120K–$160K | 0.30–0.60% |
| Growth Operations | $80K–$150K | 0.20–0.40% |
The team reads like a cap table of operator‑investors. CEO Zach Witzel ran algorithmic pricing at Uber and founded an automated marketing startup. CTO Shishir Jakati built AI at Amazon Alexa. Founding engineer Arhan Choudhury previously founded an ML ecommerce company. Head of Mobile Kyle Gorlick founded consumer mobile startups. Kenny Yinusa founded a network fraud detection startup and did Android at Goldman Sachs. Carlos Mercado and Jesús Valente Rojas both engineered at Google, with Rojas on Firebase. Cristian Dinuta, Growth, came through Google's software apprenticeship and founded an AI finance app. Trent Hommeyer, Head of Ops, ran AI operations at Accenture. Three in four team members have started a company; every new hire joins a room where the baseline is "has shipped a product from zero."
Capital followed the same pattern. The $6 million seed came from Y Combinator, Gradient (Google's AI fund), and individual leaders from OpenAI. The hiring page states the philosophy in one line: "We hire for judgment, taste, and craft — not credentials." Every role ends with a real build assignment, because the best signal is the work itself. That sentence is the thesis for everything that follows.
The Build Assignment Replaces the Resume
Assignment-based hiring replaces credential proxies with direct evidence: candidates complete a real work task, and the output becomes the primary hiring signal. Fueler.io describes the model plainly: instead of relying solely on resumes and interviews, companies evaluate the actual work produced by candidates. That shift moves the evaluation from "what does this person claim?" to "what can this person do?"
Helium AI, the platform surfaced in YouTube walkthroughs and on heliumedge.io, is built around this loop. The product lets hiring teams upload existing role descriptions and hiring guidelines, then structures the assignment flow so the work sample sits at the center of the decision. The marketing language — "still using old job descriptions that fail to attract the right candidates?" — frames the assignment as the filter that separates signal from noise. HeliumEdge, the lightweight hiring platform from the same ecosystem, positions itself as the infrastructure that makes assignment-based evaluation repeatable rather than ad hoc.
In practice, the assignment mirrors a scoped slice of the actual role. The software engineer listing describes the process: "Build assignment: pick one of three mini-tools and build it." Reviewers score against a rubric tied to the role's core competencies. The resume becomes context; the assignment becomes the verdict.
This structure also changes what candidates optimize for. When the gate is a live build task, preparation shifts from keyword-tuning a LinkedIn profile to maintaining runnable side projects, documented experiments, and public code that demonstrates the same muscles the assignment will test. The platform's design — upload guidelines, launch assignment, review output, makes that loop legible to both sides.
The careers portal reinforces the point: The software engineer listing adds a concrete tell: "A side project you can point to and say 'I built this.'" That is the credential Helium respects. Not where you studied. Not which logo sits on your resume. What you have shipped.
Are Candidates Changing How They Prepare?
Glassdoor lists three anonymous interview reviews and three interview questions for Helium, a sample too small to reveal a pattern but enough to confirm candidates are documenting their experiences. AmbitionBox similarly aggregates Helium interview questions and answers shared by applicants across roles. Neither source details whether candidates are foregrounding side projects or portfolio work in response to Helium's stated emphasis on judgment and craft.
What the broader technical-interview literature shows is a long-standing shift toward demonstrable output. Indeed's technical-interview guide frames preparation around "qualities to display": problem-solving, communication, and technical depth, rather than credential recitation. That framing aligns with what Helium's public materials describe, but the research does not capture a Helium-specific applicant cohort reorganizing their preparation around build artifacts.
The only first-hand candidate guidance in the research comes from a 2021 YouTube tutorial on virtual-interview logistics: camera height, lighting, internet speed (1.8 Mbps for Zoom), background, outfit, and distraction management. Those mechanics matter for any remote screen, but they say nothing about how applicants choose which projects to highlight or how they structure a work-sample submission.
In short, the public record shows Helium candidates are sharing interview questions on Glassdoor and AmbitionBox, and general technical-interview advice emphasizes demonstrated ability over pedigree. Whether Helium's six open roles are driving a measurable uptick in portfolio-centric applications remains undocumented in the available sources.
The Rest of AI Is Catching Up
The shift Helium exemplifies is not isolated. Across the AI sector, companies are replacing credential gates with work-sample evaluations because the old signals — degrees, brand-name employers, LeetCode scores — have decoupled from the skills that actually move models into production. LinkedIn reports that skills-based platforms such as HackerRank and Codility are displacing resume screening with coding and project challenges, while Forbes notes that AI-powered technical assessments are supplanting static coding challenges with dynamic evaluations.
The change is structural. Three in four large companies report a severe AI talent shortage. McKinsey's analysis of 4.3 million postings finds fewer than half of applicants possess the high-demand skills required. A credential filter simply discards the few candidates who can do the work.
Recruitment technology has absorbed the pressure. Equity investment in next-generation hiring software reached $17 billion in 2023, funding platforms that automate screening, scheduling, and initial outreach through generative AI chatbots. But the same tools that scale intake also expose its fragility. Startups Magazine documented a flood of ChatGPT-generated applications so templated and generic that hiring teams reject them on sight, creating a paradox where companies demand AI fluency on the job but penalize AI use in the application. The response has been to make the application itself a work sample: the test becomes the proof.
Market data reinforces the turn. AI-related job ads have jumped roughly 68 percent globally since late 2022. Postings requiring AI skills surged 61 percent year-over-year against 1.4 percent overall growth. Nearly one in four new tech roles now explicitly seeks AI skills, and generative AI postings have tripled in recent years. Yet McKinsey also found applied AI postings fell 29 percent and industrialized ML postings dropped 36 percent in 2023, evidence that hiring is concentrating in specialized, high-impact niches (NLP, multimodal systems, LLM engineering) rather than broadening. Companies are paying 20–30 percent premiums for those niches, and they cannot afford false positives.
The platforms enabling the shift are themselves evolving. HackerRank and Codility now frame challenges around realistic project scopes rather than algorithmic puzzles. Newer entrants layer model-based evaluation on top. The assessment becomes a simulation of the daily loop. Candidates who have shipped side projects, contributed to open-source model tooling, or published replication studies pass; those who optimized for interview patterns stall.
By 2025, an estimated three in four enterprises will have moved AI models into full production environments. That deadline forces hiring teams to verify production competence before offer stage. The ripple is visible in job descriptions: requirements increasingly emphasize shipped work and public repositories over degree pedigree. Helium's public rubric — judgment, taste, craft over credentials — is simply the most explicit version of a filter the market has already adopted.
What This Means for Your Next Hire
The shift Helium illustrates is already moving beyond a handful of startups. LinkedIn's hiring data shows the same pattern: when recruiters screen for proof of work, the signals that rise are portfolio, work samples, and availability, not degree pedigree or brand-name employers. That change rewrites the top of the funnel for every AI team.
Job descriptions are the first artifact to change. The same LinkedIn analysis urges hiring managers to write skills-focused descriptions that outline core capabilities, tools, and expected outcomes rather than credential checklists. In practice, that means a posting for a research engineer lists verifiable accomplishments instead of "PhD from top-10 program." The requirement becomes verifiable; the filter becomes self-selecting.
Assessment methods are following. Expertise Recruitment projects that AI-driven assessments, predictive analytics, and job simulations will become standard tools for identifying candidates on ability rather than pedigree. The logic is straightforward: a simulated debugging exercise or a take-home model-optimization task reveals judgment and craft in ways a transcript cannot. Helium's build assignment is an early, manual version of that trend; the next wave automates the delivery and scoring of similar work samples at scale.
Soft skills are gaining weight in the same motion. The Expertise Recruitment outlook notes that the shift toward skills over credentials aligns with growing emphasis on soft skills: communication, collaboration, taste, precisely the attributes Helium labels "judgment" and "taste." When the primary screen is a work sample, the review conversation naturally centers on how the candidate approached the problem, what they prioritized, and how they explain trade-offs. Those signals replace the proxy of a prestigious internship.
For talent acquisition teams, the operational implications are concrete. Sourcing budgets shift toward communities where builders publish: open-source repos, technical blogs, Kaggle competitions, and Discord servers where researchers share checkpoints. Screening time moves from resume review to work-sample evaluation. Interview loops shrink because the assignment has already answered the "can they build?" question; the remaining conversations probe judgment, communication, and cultural alignment.
Zero G Talent's job board data shows First-party hiring data reflects the volume behind the trend. Zero G Talent's job board data reports ASML added 55 roles in the past week; According to Zero G Talent's job board data, Stripe added 48. Zero G Talent's job board data indicates Both companies list salary bands that span wide ranges: Zero G Talent's job board data puts ASML $21k–$278k (median $154k), Zero G Talent's job board data's figures put Stripe $62k–$286k (median $235k), signaling that they compensate for demonstrated impact, not title hierarchy. Zero G Talent's site inventory finds When hundreds of high-leverage roles open simultaneously, the only scalable filter is evidence of work.
Working in frontier tech? Zero G Talent tracks the openings: see every open ASML role, browse frontier tech jobs, openings at Stripe, and the people building the field.