JD Haibo's team built a structured AI knowledge base using OKF specification, a three-layer architecture (project-domain matrix, role-based views, skills), and a closed-loop lifecycle with automated generation and human validation; AB experiments showed 44% faster requirement analysis and more comprehensive coverage of cross-system dependencies.
I won the Beijing Global Hackathon’s Coding Agent track solo while battling a cold, and later distilled the 48‑hour trial‑and‑error journey with Codex into an open‑source skill that captures direction‑finding, rapid tech‑stack evaluation, and trustworthy AI‑agent loop engineering.
The article traces the evolution of an e‑commerce recommendation platform through three architectural versions, detailing the scalability, modularity, and configurability challenges of V1.0, the pipeline‑based improvements of V2.0, and the dynamic configuration service, AB testing, and distributed recall and prediction services introduced in V3.0.
This article explains the fundamentals of concurrency, distinguishes connections from requests, shows how to calculate and tune maximum concurrent connections for Nginx and HAProxy, covers system resource limits, demonstrates real‑time monitoring with stub_status, and provides practical load‑testing and Prometheus monitoring guidance.
This article presents Ctrip’s novel non‑user AB testing split algorithm that combines optimized random sampling, greedy exchange, and graph‑based community detection to achieve balanced metric distribution, reduce user traffic cross‑over, and dramatically improve split efficiency in real‑world hotel marketing experiments.
AB Testinggraph community detectiongreedy optimization
This article explains the statistical foundations of AB testing, covering the Rubin causal model, SUTVA and randomization assumptions, parameter and confidence‑interval estimation, hypothesis‑testing procedures, and essential limit theorems such as the law of large numbers and the central limit theorem.
The article outlines how AA testing and rerandomization can detect and correct non‑uniform traffic splits in short‑term AB experiments, detailing three solutions—AA tests, seed‑based rerandomization, and retrospective AA analysis—along with theoretical guarantees, empirical error‑rate reductions, and remaining challenges for long‑term or clustered designs.
This article analyzes JD Retail's approach to automating AB experiment workflows by introducing a data‑weaving framework that unifies metric definitions, streamlines logical data modeling, and enables scalable, real‑time DAG orchestration across multiple experiment scenarios.
The article reviews common AB‑testing interview questions for data‑science candidates, explaining the role of p‑values, Type I/II errors, the difference between statistical and business significance, why effects can vanish when scaling, and how to improve experiment sensitivity through larger samples, variance‑reduction methods, and careful metric design.
This article presents a comprehensive overview of JD Retail's data‑weaving approach to AB experiment automation, detailing the challenges of consistency, scientific rigor, and timeliness, the logical data platform architecture, key technologies, metric modeling, automated DAG orchestration, current progress, and future directions.
Tencent Cloud AI Code Assistant has evolved from token‑level IDE completions to LLM‑driven multi‑modal coding and chat features, employing a dual‑loop R&D system, Hunyuan‑based code models, and sophisticated trigger, prompt, stop, and display strategies to deliver context‑aware, secure, and efficient code generation within IDE and review environments.
The article examines the key challenges of recommendation AB testing at Xiaohongshu—including change stability, single‑experiment precision, and multi‑strategy packaging—and presents a series of engineering and statistical solutions such as SDK‑based AB architecture, virtual PreAA experiments, CUPED/DID adjustments, and reverse experiments to improve reliability and metric impact.
This article presents a comprehensive overview of constructing and deploying a user and item profiling tag system at Qunar, covering tag taxonomy, integration challenges, technical architectures, algorithmic methods such as classification, recommendation, knowledge‑graph and causal inference, as well as real‑time streaming, ID‑mapping, and practical applications in marketing, attribution and A/B testing.
Huolala’s data team created a comprehensive metric library platform that centralizes metric definitions, classifications, data, and analysis, enabling data‑driven decision‑making, operational efficiency, service optimization, and strategic business growth across its freight services.
This article presents a comprehensive overview of building a user profile tag system—including tag taxonomy, platform architecture, construction methods, update cycles, access patterns, common algorithmic tags, and real‑world applications such as marketing, metric attribution, and A/B testing—illustrated with examples and a detailed Q&A session from a data‑mining senior manager at Qunar.
Bilibili's offline package solution accelerates H5 page load by pre‑downloading resources, intercepting WebView requests, offering snapshot, AB testing, QR‑code debugging, version control, and off‑peak releases; deployed across many projects, it cuts load times by roughly 20‑30% and improves first‑contentful‑paint.
This article explains how ByteDance integrates data‑driven A/B testing throughout its R&D workflow—from feature design and large‑scale refactoring to bug fixes, release safety, SQL optimization, and cultural adoption—demonstrating the ROI and sustainable practices of a data‑centric development culture.
This article examines how game AB experiments can exhibit significant regional differences, outlines a meta‑analysis framework to detect heterogeneity, decomposes its sources into treatment‑effect and distributional factors, and demonstrates how to predict outcomes for unseen regions using machine‑learning models.
This article details the design, integration, and operational practices of a comprehensive user and item profiling tag system, covering tag taxonomy, construction methods, update cycles, access strategies, algorithmic implementations, and real‑world applications such as marketing, attribution analysis, and A/B testing.
This article explains the limitations of conventional A/B testing in freight two‑sided markets, introduces offline spatiotemporal splitting to isolate treatment and control groups, discusses the bias‑variance trade‑off, and provides a step‑by‑step design process with practical risk considerations.
AB testing is a statistical method for comparing multiple strategies or versions to determine the most effective one, and this article explains its risks, mitigation measures, advantages, key dimensions, step‑by‑step workflow, fairness considerations, data‑cleaning techniques, target goals, design guidelines, and alternative experimental approaches.
This article explains why traditional hypothesis‑testing methods can mis‑estimate variance when the splitting unit and analysis unit differ in AB experiments, introduces the Delta Method as an unbiased variance estimator, compares it with Bootstrap and other corrections through simulations and real‑world case studies, and highlights its computational efficiency.
This article details NetEase Cloud Music's engineering solution for gradually upgrading its iOS React Native version using a dual dynamic‑library gray‑release strategy, covering background, challenges, symbol handling, code modifications, and practical issues to enable zero‑impact, data‑driven rollouts.
This article examines how time‑slice (time‑slot) AB experiments can cause capacity competition, analyzes the resulting bias‑variance trade‑off, and provides practical guidelines for selecting slice lengths and rotation methods to ensure reliable quantitative results while preserving qualitative conclusions.
This article explains how pre‑experiment bias can distort AB test results and introduces a suite of techniques—including AA retrospective analysis, SeedFinder optimal random grouping, variance reduction, and an offline splitting algorithm—to create homogeneous test groups and improve experiment reliability.
This article explains how the Huolala data‑science team tackles AB‑testing challenges, pre‑experiment differences, observational (non‑AB) studies, and advanced causal‑inference methods such as CACE, heterogeneous treatment effects, mediation modeling, regression discontinuity, and instrumental variables to derive reliable business insights.
AB TestingCausal Inferenceheterogeneous treatment effect
The 2023 Meituan Technical roundup highlights the ten most‑read articles, ranging from the evolution of its code‑hosting platform and automatic SDK generation to risk‑visualization, trace‑ID recovery, trustworthy A/B testing, KDD research, UI intent detection, MJDK compression gains, large‑scale database high‑availability, and interactive food‑delivery recommendation, each with brief summaries and links.
The paper defines AB increment, shows how to calculate DAU gains from per‑user visit rates, explains how social viral experiments introduce unidirectional or bidirectional contamination that biases increment estimates, and proposes four probability‑estimation schemes—exponential smoothing, expansion coefficients, and homogeneous‑group sampling—to correct the bias based on experiment design and business context.
This article explains a chi‑square‑based method for detecting and removing extreme‑value orders in freight AB experiments to preserve group homogeneity, improve metric reliability, and enhance the scientific validity of experimental conclusions.
This article explains Douyin’s growth‑analysis methodology, how to construct a comprehensive growth‑metric system with North‑Star indicators and hierarchical metric layers, the end‑to‑end analysis loop, new scenario‑driven metric applications, and a detailed case study on improving video‑submission rates.
This presentation by senior data mining manager Zhou Yuanwei of Qunar outlines the architecture of an image tagging platform, the construction of a comprehensive tagging system, common algorithmic tags, and real-world applications such as look‑alike marketing, A/B test efficiency analysis, and business attribution, helping audiences understand tag types, design considerations, and value‑driven use cases.
This article examines how two‑sided platforms such as Uber, Lyft, DoorDash and Didi design and evaluate peak‑surcharge experiments, addressing network effects, bias‑variance trade‑offs, time‑space slicing, random‑saturation designs, and continuous bandit‑based testing within an operations‑focused experimental system.
Huya’s data‑driven experiment platform showcases how causal inference, AB testing, and uplift modeling are applied to advertising, user activation, and growth scenarios, detailing platform evolution, metric design, statistical challenges, and practical solutions such as multi‑test correction, CUPED, RTA, and propensity‑score methods.
This article presents a comprehensive AB testing framework tailored for new‑user growth scenarios, detailing the challenges of early traffic allocation, the scientific validation of a new experiment system, real‑world case studies, and practical guidelines for evaluation and implementation.
This article explores how freight platforms can optimize transaction strategies through AB experiments, detailing common challenges such as split‑testing interference, SUTVA violations, capacity competition, homogeneity issues, and Simpson's paradox, and presents practical solutions like time‑slice routing, city isolation, and advanced statistical corrections.
This article explains how to guarantee homogeneity in AB experiments by defining pre‑experiment bias, presenting statistical testing methods, outlining a three‑step workflow for both pre‑ and post‑experiment phases, and sharing real‑world case studies and correction techniques to improve decision‑making reliability.
This article examines how leading two‑sided platforms such as Uber, Lyft, and DoorDash design and run scientific experiments—ranging from time‑space slice A/B tests to random‑saturation and continuous bandit trials—to accurately measure and improve surge‑pricing strategies despite network‑effect biases.
This article explains how Volcano Engine identified the tangled authorization, data‑fetching, and performance problems of its advertising AB‑testing platform and refactored it by splitting services, redesigning the data model with MySQL and ClickHouse, applying DAG scheduling, time‑wheel algorithms, Domain‑Driven Design, and rigorous unit testing to achieve a more stable, extensible backend solution.
This article presents a comprehensive AB testing framework tailored for new‑user growth scenarios, detailing the challenges of early traffic splitting, the design of a scientifically validated experiment system, ID selection criteria, and real‑world case studies that demonstrate improved retention and device activation.
This article explains how video on‑demand cost is dominated by CDN bandwidth, how player‑side optimizations—such as reducing cache waste, using dynamic buffering levels, smarter range requests, precise preloading, and resolution adjustments—can cut bandwidth by 8% and overall costs by 20% while preserving user experience.
This article explains how LinkedIn’s advertising teams prevent misleading AB‑test results, describes the challenges of large‑scale ad experiments such as cannibalization, reviews industry solutions, and introduces their innovative budget‑splitting experiment that dramatically improves statistical power.
This article explains how internet growth analysis combines strategic and product‑level metrics, introduces core profit formulas, critiques traditional ROI calculations, demonstrates the advantages of marginal ROI based on AB experiments, and discusses the implications for advertisers and platforms in a cost‑reduction environment.
This article provides a concise overview of recommendation systems, covering their definition, core framework of recall, ranking, and re‑ranking, various recall strategies including multi‑path and vector‑based methods, similarity calculations, and practical implementation details such as AB testing and code examples.
The paper introduces a pre‑user stratification model that uses covariate‑balancing algorithms to create separate strata for distribution and revenue metrics, ensuring equal user allocation in Vivo game‑center AB tests, which reduces metric variance, improves gray‑release effectiveness, and saves significant investigation effort.
The LiveVideoStackCon 2023 session revealed how a human‑perception‑driven end‑cloud super‑resolution framework, AI‑based no‑reference video quality assessment, and rigorous AB‑testing methods can dramatically reduce video bandwidth while enhancing visual quality, illustrating the broader challenges and opportunities in modern audio‑video systems.
The Bilibili Game Center recommendation system combines a unified feature platform, multi‑stage recall, ranking and re‑ranking models, online services, and AB experimentation to deliver personalized game suggestions, resulting in up to 78% higher click‑through, 76% higher conversion, and substantial increases in user engagement and revenue.
The article describes how Volcano Engine's DataTester team refactored the advertising AB testing platform by splitting services, redesigning the data model with MySQL and ClickHouse, introducing DAG‑based scheduling and a time‑wheel algorithm, and applying domain‑driven design and rigorous unit testing to improve stability, scalability, and maintainability.
The article explains how Baidu App reduces its iOS package size by analyzing Link Map files and applying six code‑optimization techniques—unused class and module slimming, unused method pruning with LLVM, duplicate‑code detection, utility‑method consolidation, and A/B‑test hardening—yielding up to 8 MB of savings.
This guide explains how to set up a gray‑release (canary) deployment using Nginx as a reverse‑proxy gateway, Docker containers for isolation, and two versions of a NestJS service, with traffic split controlled by cookies and configurable percentages.
The article introduces common release techniques—blue‑green, rolling, and canary (gray) deployments—explaining their principles, advantages, drawbacks, and practical considerations, and distinguishes them from A/B testing in modern software delivery pipelines.
The article describes how NetEase Cloud Music designed and deployed a comprehensive AB testing platform, covering system infrastructure, metric modeling, scientific experiment validation (including SRM mitigation and statistical power), and operational efficiency improvements to support rapid product iteration across multiple devices.
This article explains how uplift sensitivity prediction, Bayesian causal networks, and decision‑path construction are applied to improve insurance product, coupon, and copy recommendations on the Fliggy platform, detailing modeling approaches, evaluation metrics, and practical outcomes of the causal inference framework.
The article describes a modular traffic‑control platform for e‑commerce search and recommendation that lets operators quickly adjust strategies for emergencies, cold‑start items, and experiments, replacing costly multi‑team development with a unified operation center, service center, data hub, algorithmic PID controller, real‑time metrics, independent recall chain, and cross‑scene AB testing, while outlining future extensions.
This article explains gray (canary) release concepts, compares canary, A/B testing, and blue‑green strategies, and provides a complete Spring Cloud Gateway implementation—including custom load balancer, global filter, and configuration—so you can smoothly roll out new features while preserving system stability.
This article explains the purpose, methodology, and practical examples of AB testing, describing how randomized traffic segmentation, logging, and metric analysis enable data‑driven product decisions across various industries while also noting its widespread adoption and promotional resources.
The paper tackles low-response A/B tests by applying instrumental-variable techniques and optimized propensity-score matching, showing that IV methods recover treatment effects for compliant users and that a refined PSM pipeline dramatically boosts lift detection, turning previously non-significant results into statistically significant business insights.
AB TestingCausal InferencePropensity Score Matching
This article summarizes the motivations behind causal inference, its typical business applications such as intelligent decision‑making and prediction, the practical challenges of validation and data, and its relationship with A/B testing, machine learning, and deep learning, providing a concise overview for newcomers.
This article analyzes how NetEase Payment identified its product positioning in the gaming market by addressing virtual asset trade risks, high‑value transaction challenges, and unsuitable mainstream payment platforms, then outlines a data‑driven user journey with segmentation, targeted incentives, AB testing, and iterative refinements to boost adoption.
Vivo’s platformization strategy consolidates shared capabilities—such as content review, AB testing, version release, cloud services, points, and account systems—into reusable, standards‑driven platforms that accelerate development, improve safety and user experience, and illustrate how the company tackles complexity and rapid business iteration.
This collection highlights recent technical articles covering A/B testing sample‑size calculation, the diminishing relevance of React, TypeScript incremental compilation, the JavaScript new Function construct, practical front‑end quality improvements, a step‑by‑step guide to building a browser image‑scraping extension, and a roundup of useful online code editors.
The article presents iQIYI's data‑science‑driven approach to user interest segmentation, covering the design of weighted interest tags, their validation through blind surveys and AB‑tests, the creation of factual behavior tags, and advanced content‑based clustering methods for more precise audience targeting.
AB Testingcontent recommendationinterest clustering
This article reviews ByteDance's practical growth case where the new video recommendation product “M” used a data‑driven incentive system and extensive AB testing to improve first‑week user retention, outlining the design, implementation steps, and methods for identifying core product functions.
This article explains AB testing—from its biomedical origins and online adoption to its types, three essential components, core values of causal validation and quantitative growth, and key characteristics of pre‑evaluation and parallelism—providing a comprehensive guide for data‑driven product optimization.
This case study details how Haokan Video’s PC site was transformed with a responsive 24‑column grid, streamlined navigation, visual‑noise reduction, and feed‑style content loading, resulting in higher content exposure, improved click‑through rates, and stable user retention after a month‑long AB test.
The article explains how spurious correlations can mislead game designers and product managers, illustrates the difference between correlation and causation with real‑world examples, and offers practical methods such as focusing on passive variables, ensuring hard stratification, and using behavior‑chain matching to draw more reliable conclusions.
This article details the design, challenges, and solutions behind Prism, a self‑built front‑end monitoring platform that collects multi‑device SDK data, processes it through Kafka, Flink and ClickHouse, visualizes metrics, integrates with A/B testing, and outlines future enhancements for broader enterprise adoption.
This article details the end‑to‑end evolution of ZuanZuan's push notification platform, covering terminology, architecture, large‑scale holiday pushes, real‑time data handling, performance optimizations, multi‑channel integration, AB testing, and monitoring to achieve high throughput and reliability.
Even when individual algorithm modules show higher accuracy or revenue, the overall platform can decline due to factors like competitor encroachment, macro‑economic shifts, concept drift, overlapping marginal returns, attribution errors, and coupled A/B experiments, all of which require careful analysis and mitigation.
This article introduces causal analysis, outlines its major challenges such as correlation versus causation, confounding factors, and selection bias, explains a three‑step framework (association, intervention, counterfactual), and details how Beike applied these principles in a smart client‑management tool with rigorous A/B experiments.
This article details the three‑stage evolution of an e‑commerce recommendation framework—V1.0’s simple strategy‑factory design, V2.0’s vertical business split, and V3.0’s configurable pipeline with dynamic server‑client configuration, addressing scalability, isolation, and AB‑testing challenges.
This case study details JD.com's 11.11 home‑appliance campaign, covering project background, brand goals, consumer‑centric design thinking, visual creativity with interactive scenes, front‑end performance optimizations, AB‑testing results, and insights on turning design into measurable commercial value.
This article presents a comprehensive framework for evaluating brand effectiveness by measuring audience communication ability, applying target‑audience coverage and incremental coverage models, assessing brand awareness through online behavior and surveys, and using AB testing and propensity‑score matching to derive causal insights.
This article explains how Vivo Game Center adopted a modular architecture—combining componentization, configuration, and experiment-driven deployment—to enable rapid, flexible UI layout adjustments, unified backend processes, and data-driven A/B testing, thereby improving development efficiency and user‑targeted operations.
This article introduces the 5W2H analysis framework, demonstrates its application to a game’s declining retention problem, and walks through funnel analysis, user‑group profiling, and A/B testing to identify causes and design data‑driven solutions for player churn.
To meet rapid growth and diverse user needs, Vivo’s Game Center adopted a modular architecture that separates visual templates from data, enabling reusable UI components, configurable page assembly, and AB‑test driven experimentation, which together cut release cycles, streamline backend workflows, and boost scalability and developer productivity.
AB TestingComponent-Based ArchitectureFrontend Development
AB testing often fails because practitioners skip power analysis, peek at interim results, set unrealistic null hypotheses, randomize at inappropriate units, ignore sample‑ratio mismatches, choose misleading metrics, and fall prey to segmentation errors like Simpson’s paradox, any of which can invalidate conclusions.
This article examines how applying the CST design model, user segmentation, and psychological principles such as mental accounting and social proof can significantly improve conversion rates for a savings membership product.
The article explains how to design experiment evaluation metrics—from top‑down objectives to core, quality, and observation types—and construct an Overall Evaluation Criterion by processing, weighting, and aggregating metrics, providing a robust, scalable framework for credible AB‑test assessment and product optimization.
The article explains how the non‑parametric Bootstrap resampling method provides a practical, computationally efficient way to perform statistical inference in AB testing—especially with small samples, skewed data, or ratio metrics—by generating confidence intervals and hypothesis tests via repeated sampling, outperforming traditional approaches.
This article explains how Fish-Redux has been adopted across Xianyu’s core mobile flows, the limitations of its original static and dynamic adapters, and the architectural evolution that led to a unified FlowAdapter delivering dynamic page composition, component reuse, AB testing and improved developer productivity.
iQIYI’s AB testing platform integrates a layered traffic‑splitting architecture, real‑time SDK and API delivery, log‑replay data collection, and rigorous T‑test statistical analysis to enable fast, reliable product, algorithm, and operations experiments, exemplified by a UI redesign that boosted watch time by 17.85%.
This article analyzes Kuaishou's live‑streaming ecosystem, detailing causal‑inference frameworks, observational and experimental techniques such as DID, double machine learning, causal forests, uplift meta‑learners, and complex experiment designs like dual‑sided and time‑slice rotation to evaluate product and recommendation strategies.
The article describes a full‑traffic A/B testing framework for Xianyu that hashes seller IDs to create exclusive experiment and control groups, ensuring each seller sees only one strategy, and demonstrates that a chat‑incentive for new or churned sellers boosted chat exposure by 22 % and modestly improved overall buyer‑seller metrics without harming transaction efficiency.
This article describes how Tujia applied embedding techniques, inspired by word2vec and skip‑gram models, to build item‑to‑item similarity vectors for homestay recommendations, detailing the background challenges, the embedding solution, training methodology, evaluation results, practical improvements, and future development plans.
The article explains offline sampling for AB testing, detailing why it is needed, the main challenges of limited sample size, population heterogeneity, and non‑random interventions, and presents variance‑reduction, stratified sampling, IPW, and matching methods to address these issues.
This article details a practical smooth‑release process for a cloud‑office system, explains how to use Nginx health‑check endpoints to take instances offline, and presents three AB‑testing routing strategies—IP‑based, cookie‑based, and AB‑cluster proxy—complete with configuration examples, pros and cons, and deployment steps.
This article explains the concept, history, and various forms of online experiments such as AB testing, ABn, AA, and multivariate tests, highlighting their role in causal inference, value evaluation, risk control, and product optimization within modern internet businesses.
By adapting growth‑hacking concepts such as north‑star metrics, conversion funnels, and A/B testing to the software development lifecycle, this article proposes a data‑driven “efficiency hacker” model that visualizes demand delivery paths, classifies tasks with the RIW framework, and guides teams toward faster, more transparent project outcomes.
This article explores how growth‑hacking concepts can be adapted to software R&D, proposing a north‑star metric, visual demand‑delivery pathways, an RIW model for prioritization, and timeline‑based A/B testing to achieve more objective and transparent development efficiency.
This article details how 58.com’s AI Lab applied machine‑learning, recommendation, search, speech and NLP technologies to transform its CRM system into an intelligent opportunity distribution platform and sales voice assistant, describing the underlying models, the "Michigan" workflow, AB‑testing results and future AI‑driven enhancements.
This article describes how a mobile UI test framework was extended with Python scripts, configuration files, and an Enterprise WeChat robot to automate AB‑test validation across Android and iOS, reducing manual effort and improving efficiency by more than 50 percent.
This article explains why AB testing enhances user experience, outlines its end‑to‑end workflow, details configuration, split, and data models, describes a decoupled frontend AB testing architecture with Core and Coupler packages, and shares real case studies demonstrating measurable improvements.
This article presents the evolution, architecture, and key innovations of Bigo's real‑time computing platform—covering its migration from Spark Streaming to Flink, unified platform design, development tools, operational enhancements, and the efficiency gains achieved in business scenarios such as ETL and AB‑testing.
The iQIYI International backend team boosted global video delivery by introducing a web‑cache layer, Brotli compression, and a two‑level Caffeine‑Redis caching system with pub/sub synchronization, achieving up to 6% faster First Contentful Paint, near‑100% cache‑hit rates, and measurable revenue gains.
The paper presents an AB‑testing platform that structures experiments into scenes, buckets, layers and traffic, uses a salted‑hash based allocation and a two‑step “multi‑withdraw‑fill” algorithm to adjust percentages while preserving user‑experiment stability, and describes a lightweight, cache‑centric system architecture with staggered config reloads and safeguards against database spikes and zombie nodes.
The article explains why relying on a single metric like click‑through rate is insufficient for recommendation systems, and outlines a comprehensive, multi‑dimensional evaluation framework that combines business indicators, user behavior metrics, and algorithmic performance measures such as recall, precision, and AUC.
This article explains how Meituan’s delivery business has transitioned from data online to AI‑driven decision making by building a comprehensive, one‑stop machine learning platform that includes model management, data graph, feature store, AB testing, and a machine‑learning definition language to accelerate algorithm iteration and reduce operational costs.
This article explains the concept of gray (canary) releases, walks through installing Nginx with Lua and memcached modules, shows how to configure Nginx to route traffic based on client IP stored in memcached, and provides testing steps to verify the rollout.
This article presents a comprehensive overview of a recommendation system, covering its business background, purpose, detailed engineering architecture—including data sources, computation, storage, online learning, service and access layers—and discusses key challenges, module design, and practical reflections.