Senior Software Engineer Salaries Dropped 10% — ML Engineers Still Command Premium
New data from Archer Careers and the JobForesight AI Career Risk Index shows a stark divergence in the 2026 tech salary landscape. Senior software developers at major tech firms saw their base salaries drop 10% year-over-year. Meanwhile, machine learning engineers at the same companies averaged $213,000 in base pay with a 114-day time-to-hire — the longest and most lucrative hiring cycle in tech.
The JobForesight Index puts a name to what many developers feel: AI’s impact is happening at the task level, not the job level. “The variance inside your job is bigger than the variance between jobs,” the report notes. A senior developer who spends 40% of their time on code review (easily automated) is at higher risk than one whose tasks are 80% system design and architecture.
Why it matters: This is the most concrete data we’ve seen on the AI salary wedge. Generalist developers are getting squeezed. ML specialists and architects are thriving. The skill premium isn’t about knowing Python — it’s about knowing which tasks won’t be automated next year.
NZ angle: NZ tech salaries typically lag US by 30-40%. If the US market is already compressing generalist roles, NZ developers who haven’t specialised are doubly exposed.
SAP Launches Agentic AI Curriculum — Universities Get a Head Start on the Next Skill Wave
ERP giant SAP has announced a partnership programme with universities to prepare students for the “age of agentic AI.” Gartner predicts that by 2028, at least 15% of day-to-day work decisions will be made autonomously through agentic AI — and SAP wants graduates who know how to build, deploy, and manage AI agents in enterprise environments.
The curriculum covers agent architecture, multi-agent orchestration, safety guardrails, and the “human-in-the-loop” design patterns that will define how AI agents integrate into existing enterprise workflows.
Why it matters: SAP is one of the biggest enterprise software companies on earth. When SAP builds a university programme around a specific skill, it’s a signal that skill will be in serious demand. Agent orchestration — not just prompt engineering — is shaping up as the critical competency of the 2027-2030 job market.
Cognizant Creates “Frontier Certified Engineer” — A New Job Category for the AI Era
Cognizant announced two new professional certifications: Frontier Certified Engineer and Frontier Business Operator. These aren’t just training programmes — the company is positioning them as entirely new job categories that bridge the gap between AI capability and business execution.
The Frontier Certified Engineer focuses on deploying frontier AI models in production environments — model selection, fine-tuning, inference optimisation, monitoring, and cost management. The Frontier Business Operator focuses on identifying where AI creates business value and managing the organisational change.
Why it matters: This is the first major consulting firm to create AI-specific job titles that aren’t just “AI Engineer” or “Data Scientist.” The Frontier Certified Engineer is analogous to what a cloud architect became in 2015 — a specialist who could translate technical capability into business outcomes. If this catches on, expect every consulting firm and enterprise to follow.
Cyber Security Careers Need a Post-Worm-Paper Pivot
The University of Toronto’s demonstration of an autonomous AI worm that spreads through enterprise networks on a free open-weight model has implications for cybersecurity career paths. The worm bypasses traditional signature-based detection, adapts to its environment, and exploits known vulnerabilities at scale — all things that current security tools weren’t designed to counter.
The implication: AI-specific security skills are about to become the most valuable specialisation in cybersecurity. Traditional network security, penetration testing, and SOC analysis won’t disappear, but the premium will be on professionals who understand AI model behaviour, adversarial machine learning, and agent safety.
Why it matters: The worm paper showed that the attack surface has fundamentally changed — and with it, the skills needed to defend it. For anyone considering a cybersecurity career shift, the smart money is on AI security specialisation, not generalist certs.
🔍 THE BOTTOM LINE
The AI career market is polarising hard: generalist coders are losing ground, specialists are commanding premiums, and entirely new job categories are being invented faster than universities can teach them. For anyone in tech right now, the winning strategy is specialise or get compressed.