AI Isn’t Just Taking Jobs, It’s Creating Weird New Ones
The hosts of 'HTML All the Things' discuss how AI is creating entirely new job roles rather than simply eliminating existing ones. They examine four emerging positions — Forward Deployed Engineer, AI Generalist, Prompt/Evals Engineer, and Vibe Code Rescue Engineer — arguing these roles represent the early standardization of an AI-driven job market restructuring.
Summary
The episode opens by challenging the dominant narrative that AI is purely a job destroyer, arguing instead that companies are scrambling to create new roles for people who can make AI tools practically useful. The hosts note that CEOs across major companies, including FAANG-level organizations, are mandating AI adoption across all departments — coding, marketing, finance — but lack clarity on who should fill these new needs, leading to a wave of layoffs paired with targeted new hiring, often framed as 'reorgs for the AI era.'
The first role discussed is the Forward Deployed Engineer, a position with pre-AI roots at companies like Palantir. In the AI context, companies like OpenAI embed these engineers directly into large enterprise clients to understand internal workflows, identify where AI tools can be integrated, and bring back proprietary operational insights to the AI lab. The hosts emphasize the dual-access dynamic: the FDE has deep knowledge of the AI company's roadmap while gaining inside access to the enterprise's processes — a combination no typical hire could replicate.
The AI Generalist is described as a cross-functional technologist who moves fluidly between departments — HR, logistics, marketing, finance — identifying automation opportunities and rapidly deploying AI solutions without requiring a full engineering team. The hosts compare this to Upwork's framing of the role as a 'translator between business needs and AI capabilities.' They caution that job postings for this role may be vague or exploitable due to the inherently broad nature of the position, advising candidates to vet opportunities carefully.
The third category covers Prompt Engineer, Context Engineer, and Evals Engineer. While prompt engineering (crafting language inputs to reliably elicit desired LLM outputs) has been around for a few years, the hosts argue that Evals Engineering is the more technically demanding and underappreciated frontier. They describe the enormous difficulty of consistently benchmarking AI pipeline outputs — a process that is expensive, time-consuming, inherently indeterministic, and lacks any standardized framework. The hosts suggest QA engineers are well-positioned to pivot into this space.
The final role, informally called the Vibe Code Rescue Engineer, addresses the growing reality that non-technical founders are using AI coding tools to build proof-of-concept apps riddled with security holes, dead code, and scalability problems. Once these apps gain traction or investment, experienced engineers are brought in to audit and often rewrite them. The hosts frame this as a net-positive democratization of entrepreneurship — more ideas get built and tested, even if engineering teams are smaller — and note that freelancers are already monetizing this niche. They close by inviting listeners to share additional emerging AI job titles they've encountered.
About this episode
AI is changing the job market in ways that go far beyond simple automation. While some tasks are becoming easier or disappearing altogether, new careers are emerging to help businesses implement, manage, evaluate, and even repair AI-powered systems. In this episode Matt and Mike discuss the rise of roles like Forward Deployed Engineers, AI Generalists, Prompt & Evals Engineers, and ”Vibe Code Rescue Engineers,” exploring why companies are hiring for these positions and what they reveal about the future of work. If AI continues to reshape the tech industry, understanding where new opportunities are forming may be just as important as understanding what is being automated away.
Key Insights
- The hosts argue that corporate layoffs framed as AI-related are often deliberate reorgs where companies prefer to replace existing staff with 'AI-pilled' new hires rather than retrain current employees.
- Forward Deployed Engineers existed before AI — Palantir is cited as an early adopter — but the role has grown in importance because AI systems are highly indeterministic and difficult to map onto enterprise workflows without insider knowledge.
- The hosts argue that OpenAI's Forward Deployed Engineers provide a dual advantage: they carry knowledge of upcoming model capabilities while extracting proprietary workflow intelligence from enterprise clients, enabling targeted product development.
- Mike argues that Evals Engineering is the hardest unsolved problem in AI development — there is no standardized framework analogous to React for web development, and every pipeline requires its own bespoke evaluation criteria.
- The hosts note that AI pipeline evaluation runs can cost hundreds to thousands of dollars per benchmark, making iterative testing economically painful and a genuine engineering bottleneck.
- The hosts claim that nearly 100% of vibe-coded apps Mike has audited contain security vulnerabilities, often introduced because non-technical builders prompt AI to bypass access controls during testing and never revert those changes.
- The AI Generalist role is described by Upwork as a 'translator between business needs and AI capabilities,' and the hosts argue legacy businesses that dismissed early LLMs like GPT-3 are now dangerously behind and need this role to catch up.
- The hosts argue that vibe coding has democratized entrepreneurship by enabling non-technical founders with deep domain knowledge — such as a plumber or property restorer — to build and validate product ideas that would previously have required costly engineering teams.
- Matt argues that an IT background may be as relevant as a formal software engineering background for Forward Deployed Engineer roles, given IT professionals' experience building custom internal tools and working directly with end users.
- The hosts contend that emerging job titles like 'Forward Deployed Engineer' may themselves be transitional — some will standardize into lasting roles while others may disappear within months as the industry matures.
- Mike argues that the vibe code rescue engineer opportunity is currently more viable as a freelance or startup role than as a formal corporate position, as the title has not yet appeared consistently on job boards.
- The hosts suggest that QA engineers are particularly well-positioned to transition into Evals Engineering, since the core skill — systematic, repeatable testing against defined benchmarks — directly maps from traditional software QA to AI pipeline validation.
Topics
Transcript
Everyone is talking about AI taking developer jobs, but there's another side of the story starting to show up on the job boards. AI is creating entirely new roles for people who can actually make these tools useful in the real world. Companies do not just need someone who can type some prompts into ChatGPT or Gemini or whatever. They need engineers who can deploy AI into customer workflows, generalists who can spot automation opportunities across a business, and builders who can turn prototypes into products, as well as experienced developers who can clean up the mess when vibe-coded apps get too big, too fragile, too insecure, too complicated. You get the message. So this week, we're taking a look…
Full transcript available for MurmurCast members
Sign Up to AccessMore from HTML All The Things - Web Development, AI, and Developer Careers
JavaScript Isn't Always the Answer: Native Browser Features Every Developer Should Know
This episode discusses how modern browsers have evolved to handle features that previously required JavaScript or Flash, such as form validation, accordions, dialogs, popovers, and responsive images. The hosts argue for using native browser APIs and semantic HTML instead of over-engineering with JavaScript, which reduces code complexity, improves performance, and enhances accessibility.
Growing Your App: From MVP to MLP
The episode discusses the distinction between MVP (Minimum Viable Product) and MLP (Minimal Lovable Product), using a golf score-tracking app as a case study. The hosts argue that building too many features due to AI-enabled development can complicate products unnecessarily, and emphasize the importance of user feedback, gut instinct, and resilience when determining what features actually make a product worth using.
What Happens When Physical Games Disappear?
The speakers discuss Sony's decision to end physical disc production for PlayStation games by 2028, exploring the cultural impact beyond ownership and preservation concerns. They debate whether eliminating physical media removes an important part of gaming culture that supports both collectors and the broader gaming industry, using an audiophile-to-consumer market analogy.
AI Safety: From Narrow AI to Superintelligence
This podcast episode explores AI safety from narrow AI through superintelligence, discussing how current AI systems are progressing toward artificial general intelligence (AGI) and potentially superintelligence (ASI). The hosts examine the risks, alignment challenges, and governance issues that arise at each stage, while acknowledging both the theoretical nature of these predictions and the need for proactive safety measures.
Web News: Consumer Electronics Are Getting Gutted
The hosts discuss how the consumer electronics market is being severely impacted by AI chip demand and component shortages, causing significant price increases across gaming consoles, PCs, and phones. They analyze the Steam Machine's pricing as an indicator of broader market trends and advise consumers to prepare for sustained high prices rather than hoping for relief.