Nowadays, a two-year-old AI startup that has booked $4 million in events for Google, Amazon, and Notion, TechCrunch reported, is hiring seven people — and its screen rejects anyone leaning on paper credentials.
Nowadays, founded in 2023 by sisters Anna Sun and Amy Yan, automates venue sourcing and booking for corporate events. Sun, the CEO, built product and engineering at Datadog, DoorDash, and Amazon after MIT. Yan, the COO, ran strategy and operations at Google, Meta, and McKinsey after Johns Hopkins. A $2 million party seed round , TechCrunch's data shows, (no lead) closed in December 2024 from Y Combinator, Basis Set Ventures, Hike VC, VentureUs, Underdog Labs, Decacorn Capital, SBXi, E14, and dozens of angels. Six months after launch, booked business hit $2 million, TechCrunch reported. The team numbers roughly 15 in San Francisco, and active job postings have more than doubled year over year.
| Role | Salary Range | Equity Range | Experience |
|---|---|---|---|
| Forward-Deployed Engineer | $130K–$170K | 0.50%–1.50% | 3+ years |
| Founding Software Engineer | $130K–$170K | 0.50%–1.50% | 3+ years |
| Founding Product Engineer | $130K–$170K | 0.50%–1.50% | 3+ years |
| Founding AI Engineer | $130K–$170K | 0.50%–2.00% | Any (new grads ok) |
| Founding Account Executive | $80K–$240K | 0.10%–0.50% | Any (new grads ok) |
| Founding Product Designer | $100K–$150K | 0.25%–1.25% | Any (new grads ok) |
| Founding Operator | $90K–$130K | 0.25%–1.00% | Any (new grads ok) |
The board's salary band runs $86K–$198K, median $170K, Zero G Talent's board data shows. Four roles welcome new grads; three demand three-plus years. Equity tops 2 percent for AI and product engineering.
The postings emphasize ownership over pedigree. Verbs dominate the careers page: build customer-driven features, own the product cycle from conversation to production, work enterprise accounts to build custom features, turn tribal knowledge into documentation anyone can use. The Founding Operator owns the event ops flow from request to wrap. The Founding Account Executive owns the full sales cycle and partners founders on go-to-market strategy.
The public job descriptions don't spell out screening mechanics — no take-home prompt, no coding challenge, no rubric. The pattern is public: every engineering and product role carries a "Founding" or "Forward-Deployed" title, signaling that first hires shape product, not execute tickets.
The company's LinkedIn profile says: "AI automates the most time-consuming parts of planning, while our team ensures every detail is handled end to end."
AI-Generated Resumes Have Broken the Conventional Screen
Nearly half of workers (47 percent) have used AI to apply to jobs, resume writing the top use case, a CNBC-SurveyMonkey survey of 1,686 workers found in July. The volume broke the old filter. Pearson CEO Omar Abbosh told CNBC: "AI has killed the CV or the resume because what's happening is kids right now are firing off resumes to 10,000 companies built by AI and those resumes being read by AI. So you've got bots talking to bots and so that system is broken down." The CV was built for a pre-AI world. The evidence log is built for what comes next.
The flattening is measurable. Glassdoor chief economist Daniel Zhao told CNBC: "Using AI to do what everyone else does won't help you stand out. You get résumés nearly identical because people use the same prompts." The same "perfect" phrasing appears everywhere — a sanitized, algorithm-friendly version of every candidate. If everyone is optimized, no one is unique. AI cleans applications. It cannot give you a point of view, prove your impact, or connect your story to a company's actual problem.
Recruiters are reacting. Zapier recruiting leader Bonnie Dilber told CNBC that a "copy-paste" resume from AI tools can raise fraud alerts that a recruiter then has to investigate. LinkedIn verification checks have increased. Cover letters are dying: businesses "care a lot less" about them and may stop asking, especially at early-stage startups, said Julia Stiglitz, Handshake's chief education and workforce officer. Job-search success has dropped across the board.
The backlash produced a new heuristic: proof over polish. "Never use AI to write your entire résumé," said ZipRecruiter career expert Sam DeMase. "A high-effort application stands out in a sea of generic files," said Dilber. Experts suggest a different play: apply only where you're overwhelmingly qualified. Fewer applications. Stronger matches. Less noise. Use AI to prepare, then add the judgment, examples, and specifics that make the application unmistakably yours. Forward the résumé with a line about why this person is good. That line is the part AI can't write.
Nowadays sits inside this shift. Its seven open roles, all founding or forward-deployed in San Francisco, come from a team that screens for demonstrated technical work over résumé credentials. When intelligence becomes abundant, proof becomes scarce. The question is no longer "What is your job?" It's "What have you actually done? What can you demonstrate? What can you prove?"
Junior AI Candidates Now Face Senior-Level Barriers
PwC's 2026 Global AI Jobs Barometer analyzed over a billion job ads across 27 countries. The most AI-exposed junior roles are seven times more likely to demand traditionally senior skills than the least exposed. PwC defined those skills as appearing frequently in experienced, high-exposure postings but rarely in entry-level ones in 2019. The list now expected of junior staff: motivational leadership, team building, stakeholder management, process management, mentorship, data-driven decision-making.
The mechanism is straightforward: AI absorbs the routine work that once served as apprenticeship. "By taking on the relatively basic tasks in a job, AI leaves the more complex and expert tasks to people," the Barometer notes. Seniorised roles (10+ senior skills added) grew 35% since 2019; non-seniorised peers fell 10%. Global entry-level jobs in high-exposure occupations have flatlined.
The good news is that many junior workers will be spared years of drudgery on basic, repetitive tasks. The tough news is that those same workers need to quickly step up to demonstrate skills like leadership and strategic thinking.
PwC's US practice illustrates the shift: entry-level hiring cut by a third over three years, entry-level offices cut from 72 to 13, concentrated in mentorship hubs. Nearly half of CEOs expect AI to cut junior hiring; 78% worry about losing leadership skills as AI takes over training tasks.
Skill-upgrade pressure is accelerating. AI-exposed occupations evolve skills twice as fast as least-exposed roles — a 75% wider gap than last year. New tasks in AI roles are 2.5 times more likely to need empathy, judgment, creativity. Professionalised roles grow twice as fast as democratised ones (39% vs 17%) with 42% faster salary growth. The AI-skills wage premium hit 62%, up from 57% a year ago.
For Nowadays, whose seven roles all carry founding titles, the market signal is clear. The screen prioritizes demonstrated work over credentials in a market that now expects junior hires to arrive with judgment and ownership once cultivated over years. Gartner finds only 27% of organizations have a comprehensive AI strategy; 20% believe their workforce is AI-ready. Seventy-three percent of the most productive AI users are managers or executives, leaving individual contributors undertrained. The apprenticeship ladder hasn't just compressed — for high-exposure roles, it's been replaced by a demand for proof of build.
What Gets a Candidate Past Nowadays' Screen
Founders Anna Sun and Amy Yan have made it clear: the screen filters for built things, not bullet points. Job posts ask for ownership and building. That language is the map.
The parser still runs first. Parsing converts a document into structured fields (name, contact, employers, dates, titles, education, skills), and empty fields stay empty for every later step. A two-column layout interleaves titles with skills into nonsense. Headers and footers get dropped. The fix is mechanical: single column, standard section names (Experience, not "Career Journey"), dates in one format, contact details on page one, saved as text-layer PDF or .docx. That clears the deterministic gate. The probabilistic gate (the matcher using embeddings so "led a team of six" and "managed six direct reports" register as similar) rewards vocabulary alignment. Paste the Nowadays posting, mark every noun phrase naming a skill, tool, method, or credential, then rewrite your top third so the three most important terms sit in measurable sentences. If the posting says "RAG" and you wrote "retrieval-augmented generation," change it. If it says "latency reduced 40%" and you wrote "improved performance," quantify or cut.
For engineering roles, the portfolio is the closer. DG Royals found reviewers spend seconds on a résumé but minutes on a credible portfolio link. Three to six strong projects beat fifteen mediocre ones. Each project needs: a clear problem tied to a real domain (fintech, healthcare, municipal services), a working data pipeline, an evaluation of limitations, at least one deployed demo or RAG/LLM integration. Host code on GitHub with a concise README. Publish one or two LinkedIn posts explaining business impact and next production steps. Simplilearn's benchmark: good AI projects solve a real user problem, use real data, ship as a working demo. "I completed a tutorial" doesn't count.
The Account Executive and Operator roles demand a different artifact. No one expects a GitHub repo, but they do expect evidence of process ownership: a sales playbook you wrote, a workflow you automated, a metrics dashboard you built and maintained. Quantify everything — "increased quarterly pipeline by 23%," "cut onboarding time from two weeks to three days." Hiring data emphasizes precision; over-claiming ("expert in everything") flags as noise. Mirror the posting's exact verbs. If the role asks for "stakeholder management," don't write "working with clients."
Generic AI fluff gets detected. HiredAIApp reports nearly 9 in 10 hiring managers spot AI-assisted applications; more than half care. Nowadays' screen is built by people who build AI tools; they know LLM output texture. Write your own bullets. Use AI to check formatting, not generate claims. If you didn't ship it, don't list it.
Referrals still bypass the matcher. Discussions on hiring forums describe 7,000 applicants for one R&D role, 400 phone screens, 40 interviews, almost no one passing a basic skills check. A human referral puts a résumé before a reviewer who spends minutes, not seconds. Join local AI/ML groups, contribute small open-source fixes, tap alumni networks. Weak ties become referrals when a team hires fast.
The pattern across all seven roles: clear the parser with clean formatting, align vocabulary so the matcher scores you high, then hand the human reviewer a live artifact that removes doubt. The résumé opens the door. The proof of build makes them walk through it.
What This Story Leaves Out: And Why
This piece examines one startup's hiring screen as a bounded case study — Nowadays and its seven current roles. That scope is deliberate. The screen mechanics, the proof-of-build filter, the take-home and live-coding structure: that's the story. Everything else is noise.
Salary benchmarks are not this story. The board shows $86k–$198k, median $170k, all in San Francisco. Those numbers exist; they're not a market rate. Dice puts the average U.S. tech professional at $112k; AI roles command a 17.7% premium. Indeed pegs ML engineers at $175k. Tech giants pay millions. A Reddit thread shows a startup offer at $120k with near-worthless equity. None explain Nowadays' screen. They describe different populations, stages, philosophies. Conflating them produces false precision.
Remote-work policy is not this story. Every Nowadays role is listed as San Francisco; no distributed or hybrid option exists in this cycle. That's a fact about these seven openings, not the broader market. Deel counts 40+ countries with digital nomad visas. Fragomen tracks skilled visa and remote-work permits. Creators publish guides on spousal sponsorship. Those trends are real but don't change what Nowadays requires today. Candidates needing sponsorship or remote work won't find it here. That constraint belongs in the candidate's filter, not this analysis.
Visa sponsorship is not this story. The board doesn't indicate sponsorship for any role. Macro signals (Fragomen, Deel, LinkedIn) don't map to Sun and Yan's current plan. If sponsorship appears later, it's a new data point, not a variable in this screen.
Other companies' processes are not this story. Most employers use AI in hiring; most tech professionals distrust it. But Nowadays' screen is proprietary. Anthropic differs. OpenAI differs. A Series B logistics startup differs. This article doesn't claim a standard — it claims the opposite: one team's specific response, visible because roles and criteria are public. Generalizing requires evidence not provided here.
The exclusions exist for a reason. A case study only illuminates when its boundaries hold. Blur them and the signal dissolves into the same salary-survey, remote-policy, visa-guide, industry-roundup sludge this piece was written to cut through.
The screen at Nowadays doesn't ask for a résumé. It asks for a repository, a playbook, a deployed demo — something that runs. In a market where bots write the applications, the only credential that survives contact is the thing you built that still works.
Working in frontier tech? Zero G Talent tracks the openings: see every open Nowadays role, browse frontier tech jobs, the companies hiring, and the people building the field.