AUG 1, 2026
EngBrief
Search⌘K
LatestTopicsSourcesSaved
Eng&Brief

Engineering insights from the world's best tech companies, curated and summarized.

Weekly brief

Browse

TopicsSourcesFavorites

More

SearchRSS Feed
© 2026 EngBriefUpdated every 4 hours
← Sources
blog.pragmaticengineer.com icon
Engineering Management

The Pragmatic Engineer Blog

33 articles on EngBrief

The Pragmatic Engineer by Gergely Orosz covers the inner workings of Big Tech and high-growth startups. Articles dive into engineering management, system design, compensation, hiring practices, and industry trends — offering a behind-the-scenes look at how top engineering organizations operate.

Engineering ManagementSystem DesignIndustry TrendsCareer
Visit blog →

Latest Articles

The2d ago

The Pulse: Quitting Spotify Podcasts over reliability

Spotify's video podcast publishing platform has suffered from reliability issues, resulting in frequent outages and unprocessed episodes. After three incidents, including a complete platform down period, the publisher quit publishing video episodes on Spotify, citing a lack of priority on reliable operations. The decision was made after a promised incident review failed to materialize, and the publisher's experience highlighted the trade-offs between AI adoption and operational reliability.

CareerIndustry
1 min
The9d ago

The Pulse: New trend - concern about massive increase in code review load

Engineering leaders are struggling to keep up with the massive increase in code review load, particularly since AI code generators such as Opus 4.5 and GPT 5.4 started producing high-quality code. Some companies, like Uber and Cloudflare, have developed in-house AI code review tools to manage the load, while others are experimenting with and adopting dedicated AI code review tools. However, the shift to AI-generated code is also causing issues with code review quality and efficiency, with engineers feeling overwhelmed by the workload and struggling to add value through human reviews.

CareerIndustry
1 min
The16d ago

The Pulse: What can we learn from Bun’s rapid Rust rewrite with AI?

Jarred Sumner, creator of the JavaScript runtime Bun, successfully rewrote Bun's 535,496 lines of code from Zig to Rust in a matter of days using Anthropic's AI model, Fable. The rewrite aimed to address memory-safe issues in the original codebase, eliminating bugs and crashes caused by manual memory management. This rapid rewrite showcases the potential of AI-powered code generation to modernize complex software projects with unprecedented speed and efficiency.

CareerIndustry
1 min
The23d ago

The Pulse: Interesting AI coding stats from Cursor

Cursor recently released a report based on two years of data from its AI coding platform, highlighting some key trends. Top 1% of users generate 30-40K lines of code per week, equivalent to 45 median users. The report also found that input tokens (used for reading codebase and documentation) comprise 90% of Cursor's token usage, with output tokens taking up only 0.6% when caching is considered. The efficient usage of input tokens over output tokens suggests that context reuse and caching will be a crucial aspect of future AI efficiency. The report also compared the cost of different AI models, finding that Opus 4.7 is about 10x more expensive than Composer 2.5, but produces similar results.

CareerIndustry
1 min
TheJul 2, 2026

The Pulse: a new trend, smart model routing

Senior engineers at large companies want an 'intelligent' router to select the right model for tasks without overly high token costs. Several solutions currently exist, such as Factory Router, Not Diamond, and Vercel AI gateway, promising cost savings of 20-30% or more by automatically selecting models or routing requests. As demand for intelligent routing is high, especially from enterprises, vendors are likely to build or offer this functionality in the future.

CareerIndustry
1 min
TheJun 28, 2026

Pollen tried to remove my article about Callum Negus-Fancey, and Google is assisting to it

Pollen, an events tech company, attempted to remove a critical article about its founder Callum Negus-Fancey's mishandling of the business by lodging a copyright infringement claim with Google. This claim succeeded in temporarily removing the article from Google's search results, sparking concerns about the abuse of the copyright removal system. The author, the rightful copyright owner, has appealed the removal but faces challenges in addressing the potential exploitation of the system.

CareerIndustry
1 min
TheJun 23, 2026

Reliability fail: No automated zone failover for Coinbase’s global trading service

Here's a 2-3 sentence summary of the post: Coinbase suffered a 10-hour global trading outage on May 7, 2026, due to a single-availability zone (AZ) dependency in AWS, without an automated failover mechanism. Despite the high value of its operations ($5.2 trillion per year), Coinbase had no automated failover to another AZ, relying instead on manual code changes to restore trading. This lack of basic reliability infrastructure is concerning, especially compared to Coinbase's peers like Uber, which had a more robust setup in 2016.

CareerIndustry
1 min
TheJun 11, 2026

The Pulse: a trend of trying to cut back on AI spend within eng departments?

Several tech companies, including OpenAI and Uber, are grappling with high AI spending and lack of clear return on investment (ROI). As a result, many organizations are reassessing their AI budgets and considering cost-cutting measures such as token limits and model routing. Engineers are now experimenting with cheaper models for simpler tasks, leading to an increase in AI efficiency-focused sessions and a new focus on cost savings. To mitigate high AI costs, some companies are introducing measures such as token limits, "dumb" model downgrades, and smart model routing based on use case and prompt. Engineers who contribute to cost savings may be rewarded with promotions and pay rises, making it essential to translate AI cost savings to revenue generation or cost reduction.

CareerIndustry
1 min
TheJun 11, 2026

The Pulse: Antigravity 2.0 takes ‘IDE’ out of its new IDE

Google has released Antigravity 2.0, a major redesign of its AI IDE, which separates into two distinct applications: Antigravity IDE and a conversational interface similar to Codex's desktop app. However, the transition feels rushed, with confusing naming and lack of testing, leading to user frustration. The new Agent Manager concept and support for only certain models have been criticized, suggesting poor product quality and a lack of polish.

CareerIndustry
1 min
TheMay 24, 2026

The Pulse: Forward deployed engineering heats up again

Google and other companies are intensifying their demand for forward deployed engineers (FDEs), who integrate AI systems into customers' services. Google Cloud is streamlining its FDE hiring process and hiring more FDEs for its new AI-focused organization, while OpenAI and Anthropic are outsourcing FDE hiring to separate companies. FDE roles are shifting from platform engineering to more client-focused, integrator-like positions.

CareerIndustry
1 min
TheMay 20, 2026

Google Cloud deletes Australian trading fund’s infra

Google Cloud accidentally deleted UnivSuper's subscription, erasing all associated data, but the Australian retirement fund's backup on another service provider saved the day. This rare blunder occurred despite UnivSuper's replication setup across two Google Cloud regions, which was meant to preserve data in case of a regional failure. The incident is an embarrassing setback for Google Cloud, which took full blame and acknowledged UniSuper's foresight in maintaining an external backup.

CareerIndustry
1 min
TheMay 14, 2026

The Pulse: Did capacity shortages turn Anthropic hostile to devs?

Anthropic addressed capacity shortages by renting a data center from Elon Musk's SpaceX/xAI, doubling the compute limit for its Claude Pro subscribers, and removing usage restrictions on its AI model. This deal may benefit Anthropic's competitive position against rival OpenAI, which Musk is suing. Meanwhile, Anthropic's CEO has a history of publicly disagreeing with OpenAI, suggesting potential AI industry dynamics at play.

CareerIndustry
1 min
TheMay 12, 2026

TechPays has been acquired by Levels.fyi

Zsombor and author of The Pragmatic Engineer have sold TechPays, a leading tech salary site in Europe, to Levels.fyi. The acquisition aims to provide TechPays with the care and resources it needs, while also integrating European compensation data into Levels.fyi's global pay transparency platform. Levels.fyi will retain the existing TechPays data, ensuring developers can continue to access and utilize it for better compensation negotiation.

CareerIndustry
1 min
TheMay 7, 2026

The Pulse: AI load breaks GitHub – why not other vendors?

GitHub's reliability has significantly decreased, with multiple outages and data integrity issues in recent months. A data integrity incident occurred due to a bug that caused incorrect merge commits when using the squash merge method, impacting 2,092 pull requests and requiring customers to manually recover lost commits. GitHub's CTO attributed the reliability woes to a load spike from AI agent fuelled requests, which they are struggling to handle, despite a modest 3.5x load increase over two years.

CareerIndustry
1 min
TheApr 30, 2026

The Pulse: token spend breaks budgets – what next?

Large companies are struggling with the increasing cost of tokens, with one company seeing a 10x increase in token spend over six months. To address this issue, some companies are implementing limits on token spend or encouraging developers to use cheaper models such as Claude Sonnet. Others are exploring alternative solutions, such as pooled spend models, to manage costs. In mid-sized companies, some are taking a more proactive approach to managing AI spend by implementing model routing and measuring the impact of token usage on productivity and outcomes. For instance, a SaaS company reduced its costs by 30% by changing the default model, while a finance industry company is working to block expensive models and implement a pooled spend model to reduce costs.

CareerIndustry
1 min
TheApr 23, 2026

The Pulse: ‘Tokenmaxxing’ as a weird new trend

Meta and other tech companies like Microsoft and Salesforce created internal token leaderboards to track AI usage, resulting in a phenomenon called "tokenmaxxing" where employees excessively use tokens to climb the rankings. However, this approach led to massive waste, outages, and unnecessary spending, totaling $100M+ at Meta. After backlash, Meta abolished its leaderboard, while Microsoft and Salesforce continued to incentivize excessive token usage with metrics and targets. To avoid tokenmaxxing, Shopify took a different approach, promoting a more balanced use of AI tools, focusing on productivity and value added, rather than just token usage metrics. This allowed developers to use AI tools effectively, without inflating token counts for the sake of appearances.

CareerIndustry
1 min
TheApr 3, 2026

The Pulse: is GitHub still best for AI-native development?

GitHub's recent reliability issues, with an average of 3 days of issues per month, are impacting its ability to handle the increased load from AI agents generating code. The company's CTO attributed several outages to infrastructure overload, including a database cluster saturation incident and a Redis cluster configuration issue after a failover. A startup, Pierre Computer, has built an "AI-native" solution for AI agents pushing code, achieving sustained peaks of over 15,000 new repositories per minute, far exceeding GitHub's capabilities.

CareerIndustry
1 min
TheMar 27, 2026

Is the FDE role becoming less desirable?

Senior software engineers are increasingly less interested in FDE (Feature Delivery Engineer) roles due to their limited software engineering work and demands on the role as a "sales engineer" or "solutions consultant". Many developers who take FDE roles are disappointed to find the work does not align with their expectations, leading to high turnover rates. This trend suggests companies may struggle to fill FDE positions despite growing demand, making the role less desirable among developers.

CareerIndustry
1 min
TheMar 5, 2026

The Pulse: Cloudflare rewrites Next.js as AI rewrites commercial open source

An engineer at Cloudflare rewrote most of Vercel’s Next.js in one week with AI agents. It looks like a sign of how AI will disrupt existing moats and business...

CareerIndustry
1 min
TheJan 29, 2026

I replaced a $120/year micro-SaaS in 20 minutes with LLM-generated code

I used to pay $120/year for a SaaS that hasn’t added new features in four years, and didn’t fix its broken billing system for three years. Using an LLM, I...

CareerIndustry
1 min