The vLLM 0.20.1 patch, released shortly after 0.20.0, consolidates stability fixes and performance optimizations for DeepSeek V4, adds several bug fixes, updates installation instructions, and provides targeted upgrade recommendations for different user scenarios.
The article examines how a model defines an agent’s ultimate capabilities while the harness determines its production reliability, detailing continuous evaluation, context‑budgeting, tool‑error classification, multi‑model migration, and SRE‑style engineering practices needed to keep AI coding agents stable and performant.
This comprehensive guide explains why enterprises should fine‑tune large language models instead of using raw APIs or RAG, compares six fine‑tuning techniques (Full, LoRA, QLoRA, AdaLoRA, DoRA, Prompt‑Tuning), evaluates popular toolchains, outlines a step‑by‑step workflow, presents cost analyses, real‑world case studies, and practical best‑practice recommendations for 2026.
The article walks through model training, validation, ensemble learning, and deployment from an AI product manager’s viewpoint, using a churn‑prediction case to illustrate decision boundaries, metric choices, industry‑specific algorithm trade‑offs, cost considerations, and practical serving options.
ml‑intern, an open‑source AI agent from Hugging Face, automates the full ML workflow—reading papers, generating code, training and deploying models—using an asynchronous event‑driven loop with submission and event queues, supporting interactive and headless modes, Slack notifications, and multiple LLM back‑ends.
This tutorial explains the three-stage evolution from Prompt Engineering to Context Engineering and finally Harness Engineering, detailing their motivations, core components, practical implementations, and why stable, end‑to‑end AI agents require a full harness to manage tasks, context, tools, execution, state, and error recovery.
This guide walks you through deploying Alibaba’s open‑source Qwen‑3 8B model on the SumW platform using vLLM, covering environment activation, server launch with OpenAI‑compatible parameters, SSH tunneling for remote access, and Python client calls, while highlighting key configuration tips and common pitfalls.
This article presents a comprehensive, step‑by‑step framework for turning laboratory AI models into production‑ready systems, covering capability mapping, technology stack choices, model selection, prompt engineering, data pipelines, training strategies, and cross‑team collaboration to ensure stability, observability, and trustworthiness.
The open‑source GLM‑5.1 model, adapted to Baidu Baige's Kunlun XPU via the vLLM‑Kunlun Plugin, delivers record‑breaking SWE‑bench scores, eight‑hour autonomous coding, long‑context handling up to 64K tokens, and scalable deployment across tens of thousands of chips, showcasing end‑to‑end AI acceleration.
GLM‑5.1, the new 744 B‑parameter open‑source LLM from Zhipu, tops SWE‑Bench Pro with a score of 58.4, outpacing Claude Opus, GPT‑5.4 and Gemini, excels at long‑duration autonomous tasks, yet shows gaps in single‑turn generation and pure mathematical reasoning.
This article explains how AI engineering has evolved from Prompt Engineering to Context Engineering and now Harness Engineering, detailing each stage's challenges, core techniques, and real‑world practices that turn large language models into reliable, long‑running production systems.
This article explains why Ollama has become popular for local LLM deployment, outlines its core features, and provides a detailed, step‑by‑step tutorial for integrating OpenClaw with Ollama—including model selection, configuration, troubleshooting common errors, and advanced tips for customization and multi‑model switching.
Timber is a multi‑pass compiler that transforms classic ML models such as XGBoost and LightGBM into zero‑dependency C99 binaries, offering microsecond‑level inference latency, HTTP‑compatible serving, and substantial performance gains over Python runtimes, making it ideal for high‑throughput, low‑latency production scenarios.
This guide explains how to quickly deploy the new Qwen3.5-397B-A17B open‑source large model using Alibaba Cloud Function Compute's serverless GPU service, covering model features, deployment steps, required commands, and performance benefits.
The article explains how AI-driven predictive modeling, intelligent load orchestration, and self‑healing bottleneck detection can dramatically improve performance testing efficiency, reduce defect detection time by 68% and resource consumption by 41%, while outlining practical stacks and common pitfalls.
This article analyzes YOLO26’s architecture redesign that eliminates NMS, removes DFL, introduces progressive loss balancing, STAL, and the MuSGD optimizer, achieving up to 43% faster CPU inference and simplifying deployment for edge vision tasks across detection, segmentation, classification, pose estimation, and OBB.
This article reviews the three open‑source Qwen 3.5 models—including a 35B MoE, a 122B MoE, and a 27B dense version—detailing their parameter layouts, core attention designs, context length, inference performance, hardware requirements, and provides step‑by‑step code examples for loading them with Hugging Face Transformers and vLLM.
This article explains how Baidu's new GLM-5 large model is adapted to the Kunlun P800 XPU, detailing the async reinforcement learning framework Slime, optimization techniques like INT8 quantization and tensor‑parallelism, and provides step‑by‑step deployment commands using the open‑source vLLM‑Kunlun plugin.
The vLLM‑Kunlun Plugin, jointly released by Baidu Baige and Kunlun Chip, provides a high‑performance, zero‑intrusion solution for deploying open‑source large language models on domestic Kunlun XPU hardware, includes fused operators, precision‑validation and profiling tools, and supports over twenty mainstream and multimodal models.
This article explains how to combine Python for training deep‑learning models with LabVIEW for rapid Windows‑based UI development and model deployment, showing step‑by‑step usage of LabVIEW's Python Node and array passing techniques.
This guide explains what AI engineering entails, outlines the end‑to‑end workflow from problem definition and data preparation through model development, deployment, and monitoring, and highlights the essential programming, cloud, and MLOps skills, career tracks, emerging trends, and salary outlook for aspiring AI engineers.
This article explains what MLOps is, why a robust MLOps framework is essential for businesses, outlines its core components, compares MLOps with AIOps, details the benefits of investing in MLOps, and provides a step‑by‑step guide to designing enterprise‑grade AI MLOps infrastructure.
This guide explains what model deployment is, why it matters, the various deployment types, readiness criteria, best practices, common challenges, real‑world case studies, and the most popular tools and platforms for deploying machine learning models in production.
This article details how a multimodal AI model was integrated to detect and improve ID card photo quality, covering common image issues, differences between OCR and multimodal extraction, deployment strategies, performance metrics, cost estimation, and the resulting business and technical benefits.
This step‑by‑step tutorial shows how to quickly launch OpenAI’s open‑source gpt‑oss‑20b model on Alibaba Cloud PAI without writing code, configure the deployment, and start chatting with the model using the Cherry Studio client.
This guide walks you through a Bash script that automatically checks for Ollama, installs it if missing, lets you choose a DeepSeek‑r1 model size, starts the Ollama service, and runs the selected model locally, complete with usage examples and a token‑cost note.
LongCat‑Flash‑Chat, a 560‑billion‑parameter Mixture‑of‑Experts model with Zero‑Computation Experts, delivers top‑tier benchmark scores and fast inference while activating only a fraction of its parameters, and is fully open‑sourced with easy deployment scripts.
This guide explains how to deploy the open‑source Qwen3‑Coder model on Alibaba Cloud's PAI platform, use the interactive PAI‑DSW environment, run code snippets, and generate notebook tutorials with the model's agentic CLI, covering both enterprise and individual developer scenarios.
This tutorial walks developers through the complete workflow of building a house‑price regression model—from problem definition, data collection and preprocessing, feature engineering, and model selection, to training, hyper‑parameter tuning, evaluation, optimization, deployment as a Flask service, and ongoing monitoring—using Python, pandas, scikit‑learn, and visualisation libraries.
This article walks through uploading a MiniLM model to the Modelers.cn community, explains why testing is essential, demonstrates both usability and local tests with openMind, and provides complete Python code for classification and simple question‑answering, enabling developers to quickly deploy and evaluate MiniLM in practice.
This article provides a comprehensive overview of large language models (LLMs), covering their transformer architecture, parameter counts, GPU memory and storage requirements, and detailed fine‑tuning methods such as prompt engineering, data construction, LoRA, PEFT, RLHF, and DPO, along with practical deployment and inference acceleration strategies.
The article details how edge intelligence is applied to the Vivo official app to improve product recommendation on the smart‑hardware floor by abstracting the problem, designing feature engineering pipelines, training TensorFlow models, converting them to TFLite, and deploying inference on mobile devices, while also covering monitoring and performance considerations.
This article provides a comprehensive guide to large language model fine‑tuning, covering model architecture, parameter and memory calculations, prompt engineering, data construction, LoRA and PEFT techniques, reinforcement learning methods such as DPO, and practical deployment workflows on internal platforms.
This article details how Vivo’s official app implements edge intelligence to dynamically rank and recommend hardware products on its homepage, covering problem abstraction, data collection, feature engineering, model design, TensorFlow‑Lite conversion, on‑device inference, and monitoring for a personalized user experience.
This guide explains how to use Baidu Baige's AI heterogeneous computing platform to deploy the eight‑model Qwen3 family—including dense and MoE variants—via a one‑click process, covering resource configuration, inference acceleration options, and post‑deployment service access.
This step‑by‑step tutorial explains how to add a self‑deployed large language model to Volcengine's Edge Large Model Gateway, configure a secure calling channel, bind it to a gateway access key, and integrate the provided sample code for seamless API access.
Xinference is a powerful distributed inference framework that enables quick deployment and efficient serving of open‑source large language models via Docker or source installation, offering Web UI, CLI, and API interfaces with detailed setup, model launching, and Chatbox integration instructions.
This guide walks you through deploying the open‑source QwQ‑32B model on Alibaba Cloud Function Compute with CAP, covering required services, step‑by‑step deployment, cost notes, accessing the demo UI, interacting with the model, scaling settings, and resource cleanup.
The article introduces Spring AI as an emerging tool for Java developers, explains its background, goals, and core components such as data processing, model training, deployment and monitoring, showcases application scenarios like NLP, image processing, recommendation systems and predictive analytics, and also includes promotional offers for AI resources and community groups.
This article introduces Spring AI, a Spring ecosystem module that simplifies building, training, and deploying AI applications for Java developers, covering its background, goals, core components such as data processing, model training, deployment, practical code examples, use cases, advantages, challenges, and future outlook.
This article introduces Spring AI, explains its background, goals, core components such as data processing, model training, deployment and monitoring, showcases practical use cases like NLP, image processing and recommendation systems, and discusses its advantages, challenges, and future outlook for Java developers.
This guide introduces Alibaba Cloud’s open‑source QwQ-32B large language model, highlights its superior benchmark performance over competing models, explains its integrated agent capabilities, and provides step‑by‑step instructions for one‑click deployment via the PAI‑Model Gallery.
LLMOps, the end-to-end methodology for managing large language models, encompasses a curated set of development, deployment, monitoring, and local management tools—such as LangChain, vLLM, LangSmith, and Ollama—enabling practitioners to efficiently build, scale, and maintain AI applications.
The QwQ-32B model, released by Alibaba Cloud, delivers DeepSeek‑R1‑level results with only 32 billion parameters, offers integrated agent capabilities, is open‑source under Apache 2.0, and can be quickly deployed locally via Ollama or integrated into Java applications using Spring AI.
This article outlines practical cloud‑native deployment options for DeepSeek models, explains common engineering challenges such as traffic spikes, latency, security, quota control, and provides detailed AI‑gateway solutions—including fallback, content safety, API key management, gray‑release routing, caching, and observability—to ensure reliable large‑model applications.
This guide shows how to overcome on‑premise compute limits by registering a local Kubernetes cluster to Alibaba Cloud ACK One, provisioning ACS GPU resources, and deploying the DeepSeek‑R1 inference model with the vLLM framework through a series of concrete commands and YAML configurations.
The article investigates why DeepSeek frequently returns a “server busy” message, analyzing factors such as sudden traffic spikes, compute and bandwidth limitations, security attacks, and maintenance policies, and then offers actionable solutions including query optimization, off‑peak usage, third‑party cloud platforms, and local deployment.
This article details a step‑by‑step deployment of the DeepSeek R1 model on Huawei Ascend 910B NPUs, covering FP8‑to‑BF16 weight conversion, custom container image preparation, configuration of MindIE services, common pitfalls, and practical troubleshooting tips for large‑scale inference.
This guide walks you through two practical solutions for deploying the massive DeepSeek‑R1 model on Volcengine Cloud—one using Terraform for a quick two‑node GPU setup and another leveraging cloud‑native multi‑node distributed inference with Kubernetes, covering resource sizing, environment preparation, model download, monitoring, autoscaling, and storage acceleration.
This guide walks you through deploying the DeepSeek‑R1 large‑language‑model inference service on Alibaba Cloud ACK One registered clusters using ACS GPU compute, covering model preparation, OSS storage setup, PersistentVolume configuration, arena‑based service deployment, and verification steps with concrete commands and parameters.
The article describes an edge‑based video‑cover recommendation system for DeWu that leverages the MNN SDK and a lightweight MobileNetV3 model, performing on‑device inference with quantization and parallel processing to automatically select high‑quality covers, achieving sub‑second latency and boosting click‑through rates by up to 18 %.
This guide walks you through deploying various DeepSeek‑R1‑Distill models on Alibaba Cloud's Serverless AI platform CAP, covering supported models, deployment options (Ollama and Transformer), step‑by‑step template and model‑service setups, validation methods, and tips for adding custom models.
The article explains that DeepSeek is an open‑source LLM that many sellers monetize by offering paid installation packages, outlines the model lineup and size options, and provides a step‑by‑step guide to install and run DeepSeek locally with Ollama and Open WebUI.
This guide shows how to start DeepSeek large‑language models on cnb.cool in just 5‑10 seconds without downloading, using a simple three‑step process that includes forking the repository, selecting a model branch, and running Ollama or Docker commands, plus options for long‑term cloud deployment.
This tutorial shows how to log into Alibaba Cloud PAI, navigate to the Model Gallery, select a DeepSeek model such as the distilled DeepSeek‑R1‑Distill‑Qwen‑7B, and deploy it with a single click using vLLM or BladeLLM, providing endpoint and token details for immediate use.
This guide walks developers through obtaining the DeepSeek‑V3 model on the Ascend community, converting weights for GPU and NPU, loading the appropriate MindIE Docker image, launching the container, and configuring service‑level parameters to achieve efficient, out‑of‑the‑box AI inference on Ascend hardware.
This guide walks you through installing Ollama, selecting the appropriate DeepSeek model, running it locally, and exploring integration options, highlighting the benefits of offline AI such as data privacy, faster performance, and zero subscription costs.
This guide walks you through quickly deploying DeepSeek series models on the Tencent Cloud TI platform, covering model selection, resource planning, step‑by‑step service creation, free online trial, API testing via built‑in tools or curl, and managing inference services for both large and compact models.
This article explains LLMOps—its goals, core activities, benefits, best practices, and how using an LLMOps platform like Dify can dramatically cut development time, simplify prompt engineering, data preparation, monitoring, and deployment of large language models.
This article introduces Alibaba Cloud's PAI Model Gallery, detailing the DeepSeek-V3 and DeepSeek‑R1 large language models, their architectures and parameters, and provides a step‑by‑step guide for one‑click deployment of these models and their distilled variants using vLLM or BladeLLM.
This article reviews the development of MLOps, introduces the emerging LMOps framework for large‑model engineering, outlines key architectural components, discusses current challenges and industry trends, and presents future directions and standardization efforts in AI operations.
This article reviews ten mainstream LLM deployment solutions—including WebLLM, LM Studio, Ollama, vLLM, LightLLM, OpenLLM, HuggingFace TGI, GPT4ALL, llama.cpp, and Triton Inference Server—detailing their technical characteristics, strengths, drawbacks, and example deployment workflows for both personal and enterprise environments.
The supply‑chain team leveraged core MLOps practices—versioning, testing, automated reproducible pipelines, deployment monitoring, and documentation—to eliminate data leakage, ensure online consistency, and accelerate model upgrades, using traffic‑replay, FAAS‑based decoupling, and approval workflows, ultimately cutting order‑fulfillment times, reducing costs, and enabling business teams to adopt reliable AI models at scale.
This tutorial walks through a complete house price prediction project, covering data collection from Kaggle, preprocessing with pandas and scikit‑learn, model training using RandomForestRegressor, evaluation, and deployment of a Flask API for real‑time predictions, providing full code examples.
PP‑ShiTuV2, a PaddleX pipeline that integrates subject detection, deep feature encoding, and vector retrieval, delivers 91 % recall@1 on AliProducts, surpasses earlier models by over 20 points, runs efficiently on GPU and CPU, and offers simple installation, quick‑start code, and full fine‑tuning support.
This article describes how Huolala built a cloud‑native AI development platform called Dolphin to overcome low model delivery efficiency and poor compute‑resource utilization, detailing its architecture, one‑stop workflow, resource‑pooling, observability, and future roadmap for scaling AI across the company.
The Meikel Studio team created an intelligent early‑screening system for malignant skin tumors on the PaddleX low‑code AI platform, which automatically captures dermatoscopic images, segments lesions with the PP‑LiteSeg model, achieves high accuracy (mIoU 0.868) and rapid inference, and offers one‑click deployment via RESTful API to improve diagnosis efficiency and support future medical‑imaging applications.
The article introduces a multi‑LoRA strategy that lets many scenario‑specific adapters share a single base LLM, dramatically cutting GPU usage and cost while preserving performance, and explains how to fine‑tune with LoRA, merge adapters, and serve them efficiently using VLLM.
This article compares the high costs and technical challenges of building a private large language model with the benefits, flexibility, and lower risk of using Retrieval‑Augmented Generation (RAG) on a general LLM, offering practical guidance for testing teams seeking AI assistance.
In this article, 58.com AI Lab senior director Zhan Kunlin explains how the company built a multi‑layer AI platform, created a vertical large‑language model called LingXi, and developed an AI Agent system with RAG capabilities to accelerate practical AI applications across various business scenarios.
This article introduces the Alibaba Cloud PAI platform for large language model (LLM) fine‑tuning, covering model‑training pipelines, performance‑cost trade‑offs, retrieval‑augmented generation, fine‑tuning methods such as full‑parameter, LoRA and QLoRA, model selection, data preparation, evaluation, and real‑world deployment examples.
This tutorial walks through deploying the Llama‑2‑7b‑hf model on Alibaba Cloud Kubernetes (ACK) using KServe, Triton Inference Server with the TensorRT‑LLM backend, covering prerequisites, model preparation, YAML configuration, PV/PVC setup, runtime creation, and troubleshooting steps.
This article examines the challenges of extracting data from complex financial reports and presents an AI‑driven solution that combines advanced layout analysis, table recognition, OCR, and large‑language‑model integration using Baidu’s PaddlePaddle low‑code platform, detailing model selection, training, performance tuning, and deployment.
This article introduces Treelite, an open‑source library that compiles XGBoost, LightGBM, and scikit‑learn tree models into optimized shared libraries, explains its branch‑prediction and comparison‑simplification techniques, and provides step‑by‑step Python examples showing significant inference speed gains across different batch sizes.
This article introduces LLMOps by defining large language model operations, explains the three stages of LLM development, details modern fine‑tuning methods such as PEFT, Adapter, Prefix, Prompt and LoRA, outlines the architecture for building LLM applications, discusses the main difficulties of agent‑based deployments, and presents practical solutions including Prompt IDE, low‑code deployment, monitoring and cost control.
This article walks through installing LM Studio on a Windows PC, downloading models from Hugging Face, using the AI Chat interface (including a Codellama‑generated Snake game), measuring resource usage, exploring the built‑in OpenAI‑compatible API, and summarizing its strengths and limitations.
This article explains the concept, motivations, and step‑by‑step workflow for fine‑tuning large language models—specifically Qwen‑14B—covering data preparation, training commands with DeepSpeed, hyper‑parameter settings, evaluation, and deployment via FastChat, all illustrated with code snippets and configuration details.
This tutorial walks through fine‑tuning Shanghai AI Lab’s open‑source InternLM models with XTuner, explaining chat‑format conventions, loading and inference (including multimodal InternLM‑XComposer), dataset preparation, configuration sections, DeepSpeed acceleration, and memory‑efficient QLoRA details for 7‑B‑parameter chat models.
This article introduces Alibaba Cloud's open‑source Qwen1.5 large language model series, highlights its multilingual, human‑preference alignment, and long‑context capabilities, and provides step‑by‑step guidance on using PAI‑QuickStart for model deployment, fine‑tuning, and Python SDK integration.
This article introduces Tencent FinTech’s AI development platform, outlining its business background and goals, the technical challenges encountered in feature engineering, model training, and inference stability, and the comprehensive solutions—including a unified feature engine, distributed training framework, optimized deployment, and future plans for large‑scale graph training and AutoML.
Google’s newly released open‑source Gemma models let developers run powerful large‑language‑model workloads on notebooks, workstations, or cloud platforms, offering competitive performance, extensive tooling, and built‑in safety measures for responsible AI deployment.
This article details how MaShang Consumer Finance leverages large language models for sales, collection, and customer service, covering company background, AI research achievements, model training infrastructure, data‑quality and compliance challenges, prompt engineering, inference acceleration, evaluation methods, and lessons learned from real‑world deployment.
This guide introduces the open‑source Mixtral‑8x7B large language model, explains its architecture and performance, and provides detailed instructions for using Alibaba Cloud PAI‑QuickStart to deploy, invoke via API or SDK, and fine‑tune the model with LoRA on Lingjun GPU resources.
This guide walks readers through using ModelScope’s SwingDeploy service to locate, configure, and instantly deploy open‑source AI models to Alibaba Cloud Function Compute, explaining the resources created, how to invoke the model via HTTP triggers, and how to optimize performance with provisioned instances, logging, and concurrency settings.
This guide explains how to quickly select a ModelScope open‑source LLM, deploy it to Alibaba Cloud Function Compute using the SwingDeploy one‑click feature, enable reserved idle billing, and evaluate the cost savings compared with traditional GPU provisioning.
This article provides a detailed examination of large language models, covering their underlying technologies, capabilities and constraints, model families, training processes, cloud and edge deployment challenges, agent architectures, and emerging trends, offering practical insights for developers, product managers, and researchers.
This article explains how LMDeploy streamlines the deployment of Hugging Face transformer models by adding online conversion, offering an OpenAI‑compatible API server, a Gradio WebUI, and 4‑bit weight‑only quantization with AWQ, providing step‑by‑step commands, code examples, and performance insights.
TestAgent is an open-source 7B test-domain LLM that delivers multi-language test-case generation, automatic assert completion, and a rapid deployment framework, offering industry-leading pass@1 scores, a ChatBot UI, and detailed setup instructions for diverse hardware environments.
The article explains prompt engineering techniques, supervised fine‑tuning of large language models, and their practical deployment in the Mobile Tmall AI shopping assistant, detailing ChatGPT’s generation steps, Transformer architecture, prompt clarity, delimiters, role‑play, few‑shot and chain‑of‑thought prompting, SFT versus pre‑training, LoRA adapters, data collection, Qwen‑14B training configuration, SDK‑based inference, and comprehensive evaluation.
The article reviews Baidu Cloud’s Qianfan 2.0 platform, detailing its expanded model catalog, dataset library, Chinese‑language enhancements, compression and speed gains, robust AI infrastructure, application templates, and end‑to‑end data‑labeling pipeline that together lower cost and accelerate large‑model adoption across industries.
This article explains how to build and deploy risk‑control models for online micro‑loans across pre‑loan, in‑loan and post‑loan stages, covering data ingestion, feature engineering, model scoring, decision flow, optimization attempts, and monitoring practices.
This article details a step‑by‑step guide to deploying the Vicuna open‑source LLM on a single server, covering model preparation, environment setup, dependency installation, GPU and CUDA configuration, inference commands, performance evaluation, and attempted fine‑tuning, while sharing practical observations and results.
This article outlines the essential concepts, data preparation, model creation, training, deployment, and verification steps for testing machine‑learning models, highlighting dataset requirements, algorithm categories, framework choices, resource considerations, and provides a sample inference request.
This guide walks you through using Meta's open‑source Llama 2 models on Alibaba Cloud's PAI platform, covering low‑code LoRA fine‑tuning, full‑parameter fine‑tuning with PAI‑DSW, and rapid WebUI deployment via PAI‑EAS, complete with step‑by‑step instructions, code snippets, and resource requirements.
The article describes Sunshine Insurance Group's Zhèngyán Large Model Open Platform, detailing its three‑layer architecture, AutoTrain tool, self‑developed LLM, smart routing, plugin marketplace, intelligent review, and how these capabilities empower insurance marketing, sales, service, and management through AI‑driven solutions.
This article presents an end‑to‑end overview of Alibaba Cloud’s Machine Learning PAI platform, detailing the three‑stage ML workflow, challenges in model development, the role of pre‑trained and open‑source models, PAI’s architecture, a hands‑on demo, and MLOps best practices for efficient model deployment.
This article details the step‑by‑step process of preparing the environment, merging weights, installing dependencies, running inference, evaluating Vicuna‑7B against other models, and attempting fine‑tuning, while highlighting performance results, encountered issues, and future work for large language model deployment.
The article analyzes the operational, stability, and cost challenges of Haro’s AI platform, explains why a serverless FaaS architecture—specifically Knative—was selected, and details the implementation steps, performance gains, and future scenarios for AI workloads.
The team embedded a full machine‑learning ranking pipeline as an Elasticsearch plug‑in—combining real‑time and offline feature stores, hot‑loadable model jars via Dragonfly, an MLeap execution engine, and a DSL for feature definition—replacing the coarse‑ranking logistic‑regression with a tree model that adds ~10 ms latency but yields a 1.2 % AB‑test lift, while maintaining high throughput, low CPU usage, and supporting future batch deep‑learning rescoring.
Tencent’s Taiji machine learning platform, a cloud‑native, distributed parameter‑server system, provides end‑to‑end MLOps for advertising by integrating data ingestion, feature engineering, model training, evaluation, deployment, and monitoring, supporting massive models up to billions of parameters while improving efficiency, scalability, and resource management.
