This week’s signals were brutally clear: companies are now publicly using AI productivity as justification for headcount cuts, while the model vendors are shipping agentic developer tools that push automation deeper into everyday workflows. Snap said AI now generates 65%+ of its new code as it cut 1,000 jobs (16%) and targeted $500M annual savings. Meanwhile OpenAI upgraded Codex for 3M+ weekly developers and expanded “computer-use” style capability, and Anthropic released Claude Opus 4.7 with the same token pricing and new efficiency levers like caching and batch processing. The “compute arms race” also escalated, with reports OpenAI will spend $20B+ on Cerebras chips over three years, while Cerebras filed for an IPO.

The Drop

1) Snap cuts 1,000 jobs (16%) and says AI generates 65%+ of new code

What happened: Snap said it will lay off ~1,000 employees (16%), close 300+ open roles, and expects $500M in annualized cost savings. Reuters reported Snap saying AI generates 65%+ of its new code. Snap had 5,261 employees as of December and guided Q1 adj “core profit” to $233M (vs ~$186.8M estimate) and revenue +12% to $1.53B.
Why it matters for hiring: “AI productivity” is now a board-level lever for headcount. Hiring will bias toward fewer, more senior builders who can ship with copilots and agents, plus platform and reliability roles that keep AI-assisted delivery stable.
Roles likely to be available (near-term supply):

• Product engineers and platform engineers

• Data and analytics talent

• Sales engineering / solutions roles tied to ops efficiency

2) OpenAI ships a major Codex update: “for (almost) everything”

What happened: OpenAI announced a big Codex update and said 3M+ developers use it weekly. Coverage notes Codex is pushing further into “operate alongside you” computer workflows and longer-running tasks, signaling more agentic usage patterns.
Why it matters for hiring: This shifts expectations for engineering output. Teams will increasingly hire for “AI leverage” (how people use tools safely and effectively) rather than raw keystrokes.
Roles likely to spike:

• AI developer productivity / internal tools engineers

• Evaluation and QA automation engineers (tests for agentic changes)

• Security and governance roles for agent tool permissions and auditability

3) Anthropic releases Claude Opus 4.7 (pricing unchanged, efficiency levers emphasized)

What happened: Anthropic released Claude Opus 4.7 (Apr 16). Pricing remains $5 per million input tokens and $25 per million output tokens, and Anthropic highlights cost savings via prompt caching (up to 90%) and batch processing (50%).
Why it matters for hiring: Models are becoming more capable and cheaper to run for repeat workflows. Hiring demand shifts toward teams who can operationalize models: tooling, evals, and safe deployment, not just “add an LLM.”
Roles likely to spike:

• Applied AI engineers with deployment + evaluation experience

• Agent orchestration and workflow engineers

• Security engineers focused on model and tool access controls

4) Compute escalates: OpenAI reportedly commits $20B+ to Cerebras, and Cerebras files for IPO

What happened: Reuters reported OpenAI agreed to pay Cerebras $20B+ over three years for servers powered by its chips, with a possible equity stake involved. Separately, WSJ reports Cerebras filed for an IPO (mid-May target) and cites inference demand as a key driver.
Why it matters for hiring: The market is funding and scaling inference-first infrastructure. Expect increased hiring in low-level performance, distributed systems, and hardware-aware ML runtime work.
Roles likely to spike:

• Inference performance engineers (latency, throughput, cost per token)

• Cluster and systems engineers

• Networking and observability engineers for large-scale inference

5) China AI capital markets move: DeepSeek fundraising talks + StepFun IPO prep

What happened: Reuters reported China’s DeepSeek is raising funds at a $10B valuation (per The Information). Reuters also reported StepFun is unwinding offshore structure to pave the way for an IPO.
Why it matters for hiring: More capital market activity generally means more hiring in product, infra, and go-to-market, plus increased cross-border competition for top AI talent.

AI Tool of the Week

Mokka (AI sourcing + screening agent)

What it does: Mokka positions itself as an AI platform that can source and screen candidates, with claims like sourcing from 850M+ profiles and posting jobs across 250+ job boards, aimed at reducing manual sourcing and first-pass screening load.
Who it’s for: Lean recruiting teams that need more top-of-funnel output without adding recruiter headcount.

Quick pilot idea (this week):

• Choose one hard role (platform, security, infra).

• Run Mokka sourcing for 30 minutes and generate a shortlist of 30 to 50.

• Run screening for the first 20 inbound applicants and compare to human screen decisions.

Metrics to track:

• Time-to-shortlist (brief to 20 viable profiles)

• Qualified response rate (for outbound)

• Screen-to-HM pass-through rate (inbound)

• Candidate drop-off at screening stage

Hiring / Interview Insight

“AI leverage” is now a hiring criterion, not a nice-to-have

When a public company says AI generates 65%+ of new code while cutting 16% of staff, the implication is unavoidable: teams will be expected to do more with fewer people.

Actionable interview upgrade (this week):
Add a 30-minute “AI leverage” station:

• Give a small real-world task (bug triage, test plan, PR review summary, or refactor strategy).

• Candidate uses their preferred AI tool live.

• You score: reasoning, validation discipline, security awareness, and output quality.

Metrics to track:

• Pass-through rate for this station vs overall onsite

• Production defect rate for new hires at 60 days

• Time-to-merge for first 3 PRs (if you measure onboarding productivity)

Funding Watch

Artemis | $70M | Cybersecurity (anti AI-powered attacks)

Fortune reports Artemis raised $70M as it emerged from stealth. Likely early hires: detection engineering, security data engineering, AI security research.

Nava | $8.3M seed | Guardrails for AI financial agents

Fortune reports Nava raised $8.3M seed. Likely early hires: product engineering, safety/controls, integrations with finance systems.

Lua | $5.8M seed | “Agent OS” for human-agent collaboration

Lua announced $5.8M seed funding. Likely early hires: product engineering, agent orchestration, enterprise rollout.

DeepSeek | raising | $10B valuation (reported)

Fundraising talks signal headcount ramps and broader China AI competition.

Quick Bytes

• Disney to eliminate about 1,000 jobs across divisions including technology and corporate functions, per Reuters.

• Rockstar Games confirms a breach tied to a third party; ShinyHunters claims responsibility and threatened leaks (ransom deadline reported as Apr 14).

• Reuters reported investors questioning OpenAI’s $852B valuation amid roadmap shifts.

• Reuters reported VCs floating offers valuing Anthropic as high as $800B (unverified by Reuters, but notable as sentiment signal).

What to do this week

1) Add an “AI leverage” station to one onsite loop

Metric: pass-through rate + 60-day defect/onboarding metrics
Goal: hire people who can ship with AI safely, not just talk about it.

2) Source aggressively from layoff pools (Snap, Disney)

Metric: response rate + time-to-shortlist
Goal: top candidates from layoff waves disappear fast.

3) Run a third-party access audit for recruiting and developer tools

Use Rockstar as a reminder: vendor and third-party tooling is a real attack surface.
Metric: number of privileged accounts reduced, MFA coverage, time-to-disable on offboarding

That’s the Drop. This week confirmed the market’s new baseline: agentic tools are becoming default, and AI efficiency is being used as a headcount lever. The teams that win will be the ones who hire for AI-leverage, keep cycles fast, and treat security and infrastructure as first-class constraints.  Stay informed, and see you next week!