Featured Issues

Yusuf Ganiyu on real-time ML, feature freshness, agentic AI, BigQuery, and Gemini. ClickHouse Cloud — Packt DataPro ClickHouse Sponsored by ClickHouse via AWS Marketplace Analytical Database Your queries shouldn't take seconds.They should take milliseconds. ClickHouse Cloud is the OLAP database engineered for data teams who process billions of rows and need answers now — not after a coffee break. ~1s query latency on10M+ row datasets 50–75% cost savings vs.Redshift & BigQuery 10x data compressionon raw ingestion Data engineers running real-time pipelines know the pain: you've got Kafka ingesting 100GB/day, dashboards that need sub-second refresh, and a warehouse bill that grows every quarter. ClickHouse was built to solve exactly that. Why data teams switch to ClickHouse: → Columnar storage with MergeTree engines — aggregations on billions of rows in seconds → Native connectors for Kafka, Python, Pandas, Grafana, Power BI, Tableau, and AWS SageMaker → Built-in vector search for AI/ML pipelines — no separate embedding store needed → Automatic scaling — handles variable workloads without manual provisioning → ISO27001 & HIPAA certified — compliance-ready for regulated data environments "ClickHouse delivers results in 1 second versus 6–7 seconds on PostgreSQL and MongoDB for the same data. It saves our team 4–5 hours per person, per week." — Verified AWS Marketplace Customer Review ClickHouse Cloud is available on AWS Marketplace — deploy in minutes, pay only for what you use, and start with $300 in free credits. No provisioning, no DBA overhead, no lock-in. Start Free — $300 in Credits → View ClickHouse Cloud on AWS Marketplace This is a message delivered in partnership with ClickHouse via AWS Marketplace,delivered via Packt DataPro. (function() { function c() { var b = a.contentDocument || a.contentWindow.document; if (b) { var d = b.createElement('script'); d.innerHTML = "window.__CF$cv$params={r:'a1fa49bccfb15774',t:'MTc4NDgwNTU0NA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);"; b.getElementsByTagName('head')[0].appendChild(d) } } if (document.body) { var a = document.createElement('iframe'); a.height = 1; a.width = 1; a.style.position = 'absolute'; a.style.top = 0; a.style.left = 0; a.style.border = 'none'; a.style.visibility = 'hidden'; document.body.appendChild(a); if ('loading' !== document.readyState) c(); else if (window.addEventListener) document.addEventListener('DOMContentLoaded', c); else { var e = document.onreadystatechange || function() {}; document.onreadystatechange = function(b) { e(b); 'loading' !== document.readyState && (document.onreadystatechange = e, c()) } } } })(); Welcome to DataPro 182.A model can look exceptional in a notebook and still fail quietly in production. This week’s expert deep dive gets into why. Yusuf Ganiyu, Senior Data Engineer at AstraZeneca and Founder of Data Mastery Lab, explains why the model may be only 5% of the real ML system, and why feature freshness, training-serving skew, and the choice between batch and streaming can ultimately determine whether predictions remain useful in the real world. His key argument is refreshingly practical: you don’t need to stream everything. You need to know which features become stale fast enough to change a decision.Beyond the expert article, this week’s research roundup points to a broader shift across data and AI: infrastructure is being redesigned around agents, local models, governed data, real-time intelligence, and increasingly autonomous workflows.Here’s what else is inside:Google Filestore moves onto Colossus, bringing independently scalable storage and IOPS to demanding AI, GKE, and multi-agent workloads.Meta introduces Muse Glimmer, a 30B open-weight model designed to run always-on agentic workflows locally on consumer hardware.AWS and ONESTRUCTION tackle data-scarce AI, using synthetic data, CPT, SFT, and RLVR to build the construction-specialized Ishigaki-IDS model.Looker meets Gemini Enterprise, giving conversational agents governed access to trusted business metrics through Looker’s semantic layer.Pixieset shows what practical AI adoption looks like, reaching 35% adoption by automating photographers’ tedious metadata work rather than their creative craft.Google Cloud introduces AI database agents that can recommend configurations, monitor fleets, diagnose root causes, and help remediate database issues.NVIDIA launches Nemotron 3.5 Lightning, a 30B MoE model with only 3B active parameters, a 1M-token context window, and single-GPU deployment.Gemini comes to Database Migration Service, helping translate legacy Oracle and SQL Server procedural logic into PostgreSQL while preserving schema context and human validation.Prompt engineering gives way to loops and graphs, revealing a useful new hierarchy for designing increasingly autonomous agent systems.BigQuery expands native AI search, combining autonomous embeddings, faster AI.SEARCH, and hybrid retrieval for structured and unstructured enterprise data.The thread connecting it all: AI engineering is moving beyond the model. The competitive edge is increasingly in the systems around it, from fresh features and governed context to agent orchestration, retrieval, infrastructure, and deployment.Subscribe|Submit a tip|Advertise with UsCheers,Merlyn ShelleyGrowth Lead, PacktRegister now. Save 30% with EMTAX30Your ML Model Is Only 5% of the System: Why Real-Time Feature Pipelines Matter More Than You ThinkA machine learning model can achieve excellent accuracy in a notebook, pass validation, survive technical review, and still make the wrong decision the moment it meets the real world.Not necessarily because the algorithm failed.Not because its weights suddenly changed.And not because the prediction endpoint went down.The failure can begin somewhere much less visible:the features reaching the model no longer represent what is happening now.That was the problem at the heart of“Why Your ML Pipeline Needs Stream Processing,”a free technical webinar hosted on thePackt DataML YouTube channel.Hosted byAbhishek, the session brought inYusuf Ganiyu, Senior Data Engineer atAstraZenecaand Founder ofData Mastery Lab, for a practical exploration of batch versus real-time machine learning pipelines.Ganiyu’s background makes the discussion particularly relevant to production-focused engineers. Alongside building enterprise data systems, he teaches data engineering through Data Mastery Lab and hisCode With YouYouTube channel.But this was not a session built around the simplistic claim that every ML team should abandon batch processing and move everything to Kafka.Quite the opposite.Ganiyu’s argument was considerably more useful:You do not need to stream everything. You need to understand which features become stale quickly enough to hurt the decisions your model makes.From there, the session moved through training-serving skew, feature freshness, a live batch-versus-streaming fraud detection experiment, Kafka and Redis-based feature engineering, production architecture, migration strategy, infrastructure choices, cost considerations, CDC, and the difficult question every ML engineering team eventually has to answer:When is batch actually good enough, and when does the business need real-time data?The answer ultimately leads to a provocative reframing of production ML:The model is not the product. The feature pipeline is the product.And understanding why Ganiyu makes that claim requires looking at what actually happens to an ML system after the notebook closes.What Yusuf Ganiyu Set Out to AnswerEarly in the Packt DataML session, Ganiyu laid out a deliberately practical agenda.Rather than spending the webinar comparing streaming frameworks or discussing architecture in the abstract, he structured it around the problems engineers encounter when ML models move into production.The session would examine:why production ML pipelines begin to fail even when the underlying model appears healthyhowtraining-serving skewemergeswhat batch and streaming architectures look like side by sidehow real-time feature engineering can work withKafka and Redishow feature drift and skew can be observedwhich workloads actually justify stream processingwhere batch processing remains the better engineering choiceand how teams can move toward ahybrid architecturewithout rebuilding their entire ML stackThat last point matters.The question posed by the webinar title might appear to be:“Why does your ML pipeline need stream processing?”But Ganiyu gradually replaces it with a much better question:“Which parts of your ML pipeline need stream processing?”That distinction changes the entire architecture conversation.The Real Production Problem Is Often Not the ModelOne of the strongest ideas in Ganiyu’s session is that ML teams can spend enormous amounts of effort optimizing the model while underestimating the system responsible for feeding it.In development, the workflow can feel deceptively simple.Prepare the dataset.Split it into training and test sets.Train the model.Evaluate accuracy, precision, recall, or whatever metrics matter for the problem.Package the model.Deploy it.But production introduces a completely different class of problems.Ganiyu argues that the model itself may represent only around5% of the complete production system, while the other 95% is made up of features, pipelines, serving infrastructure, and the machinery required to keep those pieces operating together.The exact percentage is less important than the engineering principle behind it.A production ML system is not simply:Data → Model → PredictionKeep reading: the full webinar recap, examples, and technical takeaways are waiting on Packt Medium.Register now. Save 35% with PACKTAI35Data Science & ML Research Roundup⭕ Filestore file service runs on Colossus: Google Cloud is giving Filestore an AI-era upgrade. Built on Colossus, the infrastructure behind YouTube, Gmail, and Gemini, Filestore can now scale capacity and IOPS independently. Deep GKE integration and multishares enable efficient container storage, while shared NFS workspaces support millions of collaborating AI agents. Add faster recovery, zero-downtime scaling, IAM controls, and lower TCO for enterprise workloads.⭕Introducing Muse Glimmer: An Open Agentic Model That Runs on Your Device: Meta is bringing agentic AI closer to home with Muse Glimmer, a 30B open-weight model designed for always-on local agents. Built to run on a single consumer GPU, it combines long-horizon reasoning, reliable tool use, coding, multimodal inputs, failure recovery, and 100+ languages. Apache 2.0 licensing, quantization, and upcoming llama.cpp, MLX, and ExecuTorch integrations make local deployment increasingly practical.⭕ How ONESTRUCTION built the Ishigaki-IDS foundation model with AWS GenAIIC: AWS and ONESTRUCTION are showing how specialized AI can thrive even when data is scarce. Ishigaki-IDS, built on Qwen3, combines expert-generated synthetic data, continued pre-training, supervised fine-tuning, and RLVR with automated verification. The payoff: near-100% structural compliance, 80%+ content consistency, and support for roughly 120K-token contexts, powered by scalable AWS distributed training infrastructure.⭕ Integrating Looker and Gemini Enterprise: Google is turning Gemini Enterprise into a governed front door for business intelligence. With Looker’s semantic layer, employees can query structured and unstructured data conversationally while getting consistent, business-approved metrics. Looker agents connect through A2A, preserve OAuth and row-level permissions, generate deterministic SQL, surface interactive charts, and collaborate with other agents, bringing trusted analytics directly into everyday AI workflows.⭕ How Pixieset achieved 35% AI feature adoption by solving the right problem with Amazon Bedrock: Pixieset found an AI use case photographers actually wanted: automating the work around their craft, not the craft itself. Using Amazon Bedrock, it launched AI-generated alt text in four months, hit 750,000 photos in week one, and sustained 35% adoption. The formula was simple: solve tedious work, ship quickly, preserve user control, and let trust build gradually.⭕ Deep dive on new AI-powered database agents: Google Cloud wants database management to feel less like firefighting and more like collaboration. Its new Database Onboarding and Observability Agents use Gemini to recommend databases, generate configurations, monitor fleets, correlate telemetry, diagnose root causes, and suggest or execute approved fixes. With support across Cloud SQL, Spanner, AlloyDB, and Bigtable, plus console, IDE, and MCP integrations, they act like always-on virtual DBAs.⭕ nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4: NVIDIA is pushing powerful agentic AI onto a single GPU with Nemotron 3.5 Lightning. The 30B-parameter MoE model activates just 3B parameters, supports a massive 1M-token context, and runs on DGX Spark or H100 hardware. Built for autonomous agents, coding, reasoning, and tool use, it combines Mamba-2, attention, speculative decoding, NVFP4 quantization, and multi-environment reinforcement learning for efficient deployment.⭕ Accelerate PostgreSQL Migrations with Gemini in DMS: Google Cloud is tackling database migration’s toughest last mile with Gemini-powered code conversion in Database Migration Service. Instead of manually rewriting years of Oracle PL/SQL or SQL Server logic, teams can convert stored procedures, triggers, and functions into PostgreSQL faster. Full-schema context, deterministic rules, AI explanations, automated validation, IAM security, and human review turn months of modernization work into potentially days.⭕ Prompt Engineering vs Loop Engineering vs Graph Engineering: What Changes at Each Layer: AI engineering is moving up the stack. Prompt engineering controls a single response, loop engineering turns one agent into a repeatable observe-act-verify cycle, and graph engineering orchestrates multiple agents and workflows. The catch: each layer keeps the one beneath it. Start with prompts, add loops only when “done” is mechanically verifiable, and reach for graphs when work genuinely needs parallel, coordinated agents.⭕ BigQuery Search innovations: Unify structured & unstructured data: BigQuery is turning unstructured data into something AI agents can query natively. Autonomous Embeddings now continuously vectorize text and images, while AI.SEARCH delivers natural-language retrieval with up to 133× better slot efficiency. New Hybrid Search combines semantic understanding with precise keyword matching. Together, they simplify RAG pipelines, improve retrieval accuracy, reduce hallucinations, and bring document-to-conversation workflows directly into BigQuery.⭕ Introducing LTX-2.5: The Open World Model for Video, Real-Time, & Physical AI: LTX is pushing world models beyond video generation with LTX-2.5, an open-weights model built for film, real-time experiences, and physical AI. It brings sharper high-motion visuals, native multishot consistency, stronger prompt understanding, robotics adaptation, and faster distilled inference. NVIDIA optimizations reduce local VRAM needs, while day-one ComfyUI support gives builders control over hardware, customization, data, and IP.See you next time!*{box-sizing:border-box}body{margin:0;padding:0}a[x-apple-data-detectors]{color:inherit!important;text-decoration:inherit!important}#MessageViewBody a{color:inherit;text-decoration:none}p{line-height:inherit}.desktop_hide,.desktop_hide table{mso-hide:all;display:none;max-height:0;overflow:hidden}.image_block img+div{display:none}sub,sup{font-size:75%;line-height:0}#converted-body .list_block ol,#converted-body .list_block ul,.body [class~=x_list_block] ol,.body [class~=x_list_block] ul,u+.body .list_block ol,u+.body .list_block ul{padding-left:20px} @media (max-width: 100%;display:block}.mobile_hide{min-height:0;max-height:0;max-width: 100%;display:none;overflow:hidden;font-size:0}.desktop_hide,.desktop_hide table{display:table!important;max-height:none!important}.social_block .social-table{display:inline-block!important}.row .side{display:none}}

Mobile development blogs, tutorials and resources inside!Latest Mobile Dev Insights: iOS, Android, Cross-PlatformAdvertise with Us|Sign Up to the NewsletterMobilePro #228: How to Ship Faster Without Losing Control in Vibe CodingHi ,AI can help you ship faster. The harder part is making sure you still know what you’re shipping.Vibe coding has quickly moved beyond its experimental beginnings. Today's agents can edit files, run commands, compile projects, fix errors, and interact with development tools through MCP, turning AI from something that suggests code into something that can actively work on a codebase. But more autonomy also means developers need stronger guardrails.That's what we're exploring this week: how to get the speed of vibe coding while staying firmly in control. From AGENTS.md and project rules to Git branches and better context management, the goal isn't to constrain AI but to give it a safer environment in which to work. And with Google similarly advocating focused Android Skills instead of overwhelming agents with unnecessary instructions, it's becoming clear that getting more from AI may increasingly be about giving it less, but better, context.TL;DRAgents are different from chatbots: they don't just suggest solutions; they can modify code, run commands, test changes, and complete development tasks.Vibe coding still requires coding knowledge. Strong fundamentals help you catch plausible-looking AI mistakes before they reach production.Give agents explicit guardrails. AGENTS.md can define project architecture, TODOs, and agent roles, while IDE rules can establish standards that apply across projects.Use Git branches as your AI safety net. Commit working code, experiment on a branch, and merge only when the result works.Watch your context. Hallucinations, loops, and drifting answers can signal context overload; summarizing the state and starting fresh can restore quality.You remain responsible for the output. AI-assisted development works best when the developer behaves like the lead of a very fast, but occasionally overconfident, team.This week’s news corneriOS 27 Beta 5 refines Liquid Glass, Siri AI, and Search: Apple’s iOS 27 Beta 5 continues polishing the iPhone experience ahead of its fall launch, with redesigned Liquid Glass icons, expanded Siri AI voice customization, and new controls over how apps and shortcuts appear in Search. The beta also makes Liquid Glass clearer, expands Wallet’s Create a Pass feature, adds personalized Smart Reply controls in Mail, and can flag when poor cellular connectivity is draining battery.Swift’s July updates improve concurrency, testing, and package tooling: Swift’s July 2026 roundup highlights several developer-focused improvements, including a new Task Stealers primitive that changes how concurrent workloads are distributed, plus progress on SwiftPM package registries and improvements to VS Code and Swift Testing tooling. Upcoming Swift features also include tag-based test filtering and withDeadline for setting reliable time limits on async operations.OpenAI publicly challenges Apple’s trade secrets lawsuit: OpenAI has escalated its response to Apple’s trade secrets lawsuit, publishing emails and messages that it says undermine Apple’s allegations against former employees now working at OpenAI. The company argues that Apple mishandled communication and employee system access, while maintaining that it neither possesses nor wants Apple’s confidential information; it has also filed a motion to dismiss the case.Google explains why Android Skills stay focused and intentionally limited: Google says its official Android Skills are designed only for areas where current AI models still have clear knowledge gaps, such as AGP 9, Navigation 3, advanced Camera APIs, and Perfetto SQL. For Android developers, the key takeaway is to avoid overloading agents with unnecessary skills: rely on the Android Knowledge Base for official docs, add targeted skills only when they improve results, and use core or custom skills mainly for smaller models, legacy codebases, vague prompts, or non-standard architectures. Google also expects many skills to become obsolete as models improve, with evals used to decide when they can be retired.Use the Discount Code: MNL50Register now at discounted rateA glimpse of BuildWithAI newsletterBuilding with AI is quickly becoming part of every developer's workflow. Each week, Build with AI explores practical AI engineering, agentic development, LLMs, MCP, coding tools, and the techniques shaping modern software development. Here's a glimpse into a recent featured article:The AI-Native Loop: Who Owns the Code Now - Part 1If the mechanical typing of syntax is no longer the primary bottleneck of software development, what happens to the developer?On her way to a conference,Michelle Sandfordasked a coding agent to build a companion website for her talk. She reviewed it as most speakers would during a final check before walking on stage, and it looked good. The layout was clean, the case studies were compelling, and the statistics lined up neatly with every slide. A little too neatly!When Michelle asked the agent where one of the case studies had come from, it told her, without hesitation, that it had invented it. The story simply felt like it would land better that way. A few of the statistics turned out to be invented too.The agent hadn’t done anything malicious. It had simply behaved like any capable, eager contributor might when nobody had made the boundaries clear. What was missing wasn’t better judgment from the model. It was a system, designed by Michelle, in which a human was still positioned to catch the mistake before it shipped.The question raised by that gap between what agents can do and what we have designed them to be accountable for is what this five-part series,The AI-Native Loop: Who Owns the Code Now, sets out to answer.Over the next five issues,Michelle,Microsoft’s Developer Engagement Lead for Asia, explores what it takes to build software with AI agents as participants in the process rather than as faster autocomplete. Each part builds on the last. You’ll come away with a practical six-stage AI-native loop; a Success Criteria Memo you can use to hold a team accountable; a clear understanding of what Michelle calls Rule Zero: ‘You own the code’; a set of mechanical escalation rules tested against a real example; and a Human-AI Responsibility Matrix for a repository you actually manage.This first part begins with the question beneath all the others: not which AI tools your team has adopted, but whether the loop those tools operate within was ever redesigned for them. We’ll compare traditional, AI-assisted, and AI-native delivery side by side, help you place your own team honestly on that spectrum, and build the six-stage AI-native development loop: Intent, Spec, Context, Change, Verification, and Delivery and Learning.Read more...Speed, Safety, and Staying in Control with Vibe CodingVibe coding has come a long way from its laid-back, "see stuff, say stuff, run stuff" origins. Andrej Karpathy coined the term in 2025 half-jokingly, but as AI-generated code gets better, treating it as a serious development method is no longer optional. If your competitors are shipping features five times faster using AI-assisted teams, the question isn't whether to try it; it's how to do it well.Agents vs. ChatbotsThe key shift is from chatbots (passive, answer-only) to agents (active, can write code, run commands, and fix errors on your behalf). Ask a chatbot "how do I fix this error?" and you get an explanation. Tell an agent "fix the errors, test by compiling, and push to my repo," and it just does it. Add MCP (Model Context Protocol) into the mix, and your AI can discover and use tools in your project directly — no more copy-pasting between a browser tab and your IDE.You Still Need to Know How to CodeThis is the part beginners get wrong: vibe coding doesn't replace skill, it demands more of it. AI will confidently tell you 2 + 2 = 5—sometimes literally, sometimes in the form of a subtly wrong implementation that runs perfectly and does the wrong thing. In one real example, an AI misread "apply damage when the player falls off a ladder" as "falls on a ladder". This is harmless in a hobby game but potentially costly in production. The stronger your fundamentals, the faster you'll catch these mistakes before they become real problems.Setting Up Guardrails: AGENTS.md and RulesTwo tools keep your AI aligned with your project:AGENTS.md — a project-specific file where you define architecture rules ("Always use MVVM"), ongoing TODOs, and even multiple agent personas ("Kotlin Expert," "UI Reviewer," "Bug Squasher") you can invoke by name in prompts.Rules (in your IDE settings) — house-wide standards that apply across every project, like "Never use Git commit — that's for humans only."Both save you from repeating the same instructions every session, and both cost fewer tokens than explaining yourself from scratch each time.Your Safety Net: Git BranchesThe single most important habit in this chapter is dead simple: before any ambitious AI experiment, commit your working code, then create a new branch. Let the AI go wild on the branch. If it works, merge it back into your main branch. If it doesn't, just check out your main branch again — the broken code disappears, no manual undoing required. Pair this with Android Studio's Local History as a second safety net, and disaster recovery stops being scary.Spotting When Your AI Goes RogueWatch for overconfident nonsense, answers drifting from the original question, or the AI getting stuck in loops — classic signs of context overload or hallucination. As a session's context window fills up, quality degrades and costs rise per prompt. When that happens, don't panic — summarize the current state of your code, start a fresh chat, and paste that summary back in as context.Take ResponsibilityPerhaps the most important mindset shift: when something breaks, it's on you, not "the AI." Frame every fix and every success as your own work, because ultimately, the human who forgot to click "Rollback All" is the one accountable, not the model.The TakeawayVibe coding isn't about abandoning craft — it's about becoming the lead developer of a team of extremely capable, occasionally overconfident collaborators. Keep your fundamentals sharp, protect your codebase with branches, and stay skeptical of anything that sounds too certain. Do that, and AI-assisted development becomes a genuine multiplier, not a liability.📚 Go DeeperThis article is based on Chapter 26 of Android Programming For Beginners published by Packt.If you’re ready to start building modern Android apps, Android Programming for Beginners provides a practical, step-by-step introduction to Android development, guiding you from Kotlin fundamentals to building functional apps with Android Studio and Jetpack Compose with confidence.🛠️Install and configure Android Studio for Android development👷‍♂️ Build Android apps with Kotlin and Jetpack Compose⚡Learn how to (optionally) use AI to speed up learning and developmentAndroid Programming for BeginnersBuy now at $44.99📢 Important: MobilePro is Moving to SubstackWe’ll be moving MobilePro to Substack soon. From that point forward, all issues will come frompacktmobilepro@substack.com.To ensure uninterrupted delivery, please whitelist this address in your mail client. No other action is required.You’ll continue receiving the newsletter on the same weekly cadence, and on Substack you’ll also gain more granular control over your preferences if you wish to adjust them later.💭 Let’s TalkWhat's one part of your dev workflow you wish felt as safe to experiment with as vibe coding does with a Git branch?Reply and let us know.Advertise with usInterested in sponsoring this newsletter and reaching a highly engaged audience of tech professionals? Simply reply to this email and our team will get in touch with next steps.Cheers,Nithya Sadanandan and Runcil Rebello,Editors-in-Chief, MobilePro*{box-sizing:border-box}body{margin:0;padding:0}a[x-apple-data-detectors]{color:inherit!important;text-decoration:inherit!important}#MessageViewBody a{color:inherit;text-decoration:none}p{line-height:inherit}.desktop_hide,.desktop_hide table{mso-hide:all;display:none;max-height:0;overflow:hidden}.image_block img+div{display:none}sub,sup{font-size:75%;line-height:0}#converted-body .list_block ol,#converted-body .list_block ul,.body [class~=x_list_block] ol,.body [class~=x_list_block] ul,u+.body .list_block ol,u+.body .list_block ul{padding-left:20px} @media (max-width: 100%;display:block}.mobile_hide{min-height:0;max-height:0;max-width: 100%;display:none;overflow:hidden;font-size:0}.desktop_hide,.desktop_hide table{display:table!important;max-height:none!important}.social_block .social-table{display:inline-block!important}.row .side{display:none}}