Author

Bilibili Tech

Provides introductions and tutorials on Bilibili-related technologies.

398

Articles

Likes

1.5k

Views

Comments

Latest from Bilibili Tech

100 recent articles max

Bilibili Tech

Feb 21, 2025 · Databases

Applying ClickHouse Bitmap and BSI Techniques for Real-Time Audience Selection in a Data Management Platform

By integrating ClickHouse bitmap structures, a dictionary service for dense ID mapping, and Bit‑Slice Indexes, Bilibili’s Data Management Platform now supports flexible, multi‑dimensional audience selection and profiling over petabyte‑scale data with minute‑level latency, cutting storage by over twenty‑fold and query times from hours to seconds.

BSIBig DataClickHouse

0 likes · 23 min read

Applying ClickHouse Bitmap and BSI Techniques for Real-Time Audience Selection in a Data Management Platform

Bilibili Tech

Feb 18, 2025 · Artificial Intelligence

Algorithmic Empowerment of Bilibili Streaming: VOD Transcoding Decision, Resource Estimation, and Live Comment Semantic Analysis

The article details how Bilibili leverages AI algorithms—including XGBoost, statistical rules, XDeepFM, and fine‑tuned SBERT—to optimize VOD transcoding decisions, estimate compute resources and processing time, and analyze live comments, thereby boosting streaming efficiency, utilization, and user experience.

AITranscoding OptimizationXGBoost

0 likes · 19 min read

Algorithmic Empowerment of Bilibili Streaming: VOD Transcoding Decision, Resource Estimation, and Live Comment Semantic Analysis

Bilibili Tech

Feb 14, 2025 · Artificial Intelligence

Can Label Over‑Smooth (LOS) Boost Long‑Tail Classification? New Metrics and Benchmarks Revealed

This article analyzes classifier re‑training for long‑tailed visual recognition, introduces two novel evaluation metrics—Logits Magnitude and Regularized Standard Deviation—proposes the Label Over‑Smooth (LOS) method, and demonstrates its state‑of‑the‑art performance across CIFAR‑100‑LT, ImageNet‑LT, and iNaturalist2018 datasets.

Machine Learningbenchmarklabel smoothing

0 likes · 11 min read

Can Label Over‑Smooth (LOS) Boost Long‑Tail Classification? New Metrics and Benchmarks Revealed

Bilibili Tech

Feb 11, 2025 · Artificial Intelligence

Building a Scalable AI Agent for Code Review: Practices, Architecture, and Challenges

The article outlines how to build a scalable, modular AI code‑review agent using LangChain, detailing stages from naive prompting to advanced prompt engineering, architecture with six core modules, strategies to curb hallucinations, improve reliability, performance, and human‑AI collaboration, and future RAG integration.

AI agentCode ReviewLangChain

0 likes · 22 min read

Building a Scalable AI Agent for Code Review: Practices, Architecture, and Challenges

Bilibili Tech

Feb 7, 2025 · Frontend Development

Building a WebRTC Video Call System: Signaling, Direct Connection, and Selective Forwarding

The article explains how to build a WebRTC video‑call system using standard APIs, detailing signaling via SDP exchange, direct peer connections, the transition to a selective‑forwarding server that forwards streams efficiently, and employing a data channel for RPC‑based room and stream management across web, Android, iOS, and Windows clients.

SDPSelective ForwardingSignaling

0 likes · 16 min read

Building a WebRTC Video Call System: Signaling, Direct Connection, and Selective Forwarding

Bilibili Tech

Jan 24, 2025 · Operations

Design and Implementation of a CDN Edge‑Node Scheduling System for Bilibili Live Streaming

The paper presents Bilibili’s multi‑layer CDN edge‑node scheduling system, which groups heterogeneous nodes by quality and price, uses cost‑aware and resource‑aware heuristics—including maximum‑flow regional borrowing and contextual‑bandit utilization prediction—to allocate bandwidth per business, achieving a 43 % bandwidth reuse increase, 33 % coverage boost, and markedly lower stall rates.

BilibiliCDNcost optimization

0 likes · 10 min read

Design and Implementation of a CDN Edge‑Node Scheduling System for Bilibili Live Streaming

Bilibili Tech

Jan 21, 2025 · Artificial Intelligence

Accelerating Large Model Inference: Challenges and Multi‑Level Optimization Strategies

The article outlines how exploding LLM sizes create compute, memory, and latency bottlenecks and proposes a full‑stack solution—operator fusion, high‑performance libraries, quantization, speculative decoding, sharding, contiguous batching, PageAttention, and specialized frameworks like MindIE‑LLM—to dramatically boost inference throughput and reduce latency, while highlighting future ultra‑low‑bit and heterogeneous hardware directions.

Inference AccelerationOperator fusioncontinuous batching

0 likes · 21 min read

Accelerating Large Model Inference: Challenges and Multi‑Level Optimization Strategies

Bilibili Tech

Jan 17, 2025 · Backend Development

NeighborHash: An Enhanced Batch Query Architecture for Real‑time Recommendation Systems

NeighborHash is a distributed batch‑query architecture for real‑time recommendation systems that combines a cache‑line‑optimized hash table—featuring Lodger Relocation, bidirectional cache‑aware probing, and inline‑chaining—with an NVMe‑backed key‑value service, versioned updates, and asynchronous memory‑access chaining to achieve sub‑microsecond, high‑throughput top‑N retrieval.

AMACNVMePerformance Optimization

0 likes · 20 min read

NeighborHash: An Enhanced Batch Query Architecture for Real‑time Recommendation Systems

Bilibili Tech

Jan 14, 2025 · Artificial Intelligence

Technical Practices and Productization of Intelligent Advertising Title Generation for Bilibili

We built an LLM‑powered system for Bilibili that automatically creates ad titles from user keywords, employing fluency, style, and quality classifiers, mixed domain data cleaning, and alignment methods such as SFT, DPO and KTO, resulting in a product that now generates about ten percent of daily titles and drives significant ad spend.

AI alignmentAd Title GenerationBilibili

0 likes · 24 min read

Technical Practices and Productization of Intelligent Advertising Title Generation for Bilibili

Bilibili Tech

Jan 10, 2025 · Frontend Development

Design and Evolution of Bilibili's Activity Platform Low‑Code System

Bilibili’s Activity Platform low‑code system was completely redesigned over two years, introducing a modular, decoupled architecture with an iframe‑based canvas, unified renderer, and Eva‑CLI tools that cut activity build time from three days to 0.48 days, boosted coverage to 77% of activities, and reduced first‑contentful‑paint by 35%.

BilibiliFrontend ArchitectureLow‑code

0 likes · 31 min read

Design and Evolution of Bilibili's Activity Platform Low‑Code System