A proposal: how to hire junior engineers in the age of AI
The author argues that junior engineer hiring has collapsed because the old coding interview no longer predicts who will do well at AI-assisted "product engineering." He proposes a structured, job-relevant replacement built around an LLM role-playing a customer.
- Junior software developer employment (ages 22-25) is down nearly 20% from its late 2022 peak, and Anthropic's junior share of hires fell from 27% to 9%, while senior hiring keeps growing.
- Coding interviews are being dropped or replaced by AI-assisted rounds at companies like Canva, Meta, and Google, but there is no agreed replacement yet.
- Unstructured interviews predict job performance far worse (.19) than structured ones (.42), so the new interview must be structured, with every candidate given the same task.
- His proposed task: a candidate spends an hour extracting requirements from an LLM playing the customer, then builds working software with a second LLM, testing product sense, systems intelligence, and written explanation.
Junior software hiring has collapsed since about 2023, and the author, Laurie Voss's successor-style essayist Seldo (writing on seldo.com), argues the cause is partly that nobody knows how to evaluate a junior for the new job. His earlier "We are all Product Engineers now" post, later a talk that has drawn about 250,000 views, argued that LLM-written code has shifted the work away from programming toward judgment about what to build. Stanford's "Canaries in the Coal Mine" research found employment of developers aged 22 to 25 down nearly 20% from late 2022, while senior hiring kept growing. Anthropic reportedly cut its junior share of hires from 27% to 9% since 2025, and 71% of software job-posting growth between May 2025 and May 2026 asked for seniors. The old coding interview is dying: Anthropic has reportedly dropped it on at least one team, Canva swapped its computer science fundamentals round for an AI-assisted one, Meta did the same with one of two onsite rounds, and Google is piloting similar changes.
Seldo's case for hiring juniors is mostly selfish, and he says so. Individual managers won't train the next generation of seniors because each would rather another company pay for it, but juniors are cheap and available while seniors are scarce and expensive, so a company that finds good product engineers can hire them in bulk. He rejects the usual workarounds. Paid internships work, with Shopify growing from about 100 interns a year to over 1,000 and Cloudflare targeting 1,111 in 2026, but they are slow and expensive and out of reach for startups. Hiring from elite schools just filters for family wealth. "Hunger" and "taste" are vague, and "hunger" reads as a proxy for youth, which is illegal to screen for. He also points to research by Sackett and colleagues (2022) putting the predictive validity of unstructured interviews at .19 against .42 for structured ones, and notes Google's brainteasers predicted nothing. His answer is structured interviews built around the real components of product engineering, starting with product sense: working out what problem actually needs solving from a messy request, including the assumptions the user never states.
The second component he proposes is intelligence in the specific sense of holding a whole system in your head, spotting problems, and reasoning about what changes when constraints shift, which he distinguishes from general IQ-style testing. He also wants clear writing, which he has always prized in hires and thinks matters more now that the work is mostly specifying and reviewing. The piece is the first half of a proposal, and the interview design itself is still being worked out in the portion that follows.