Author

Fei's Miscellaneous Talks

Random snippets of everyday life

Articles

Likes

Views

Comments

Latest from Fei's Miscellaneous Talks

7 recent articles

Fei's Miscellaneous Talks

Oct 1, 2026 · Fundamentals

Why Saved Resources Fail in Projects: A Recommender System Approach

The article applies recommender system concepts — content pool, feature tagging, indexing, recall, and ranking — to personal knowledge management, advising readers to annotate saved resources with specific problem contexts, build a searchable index, and use AI for summarization and retrieval so that collected materials become actionable when needed.

AI-assisted learninginformation retrievalknowledge organization

0 likes · 8 min read

Why Saved Resources Fail in Projects: A Recommender System Approach

Fei's Miscellaneous Talks

Sep 14, 2026 · Backend Development

Recommender System Re-ranking: List-Level Optimization Under Multiple Constraints

This article systematically dissects the re-ranking stage in recommender systems, explaining how hard constraints (filtering, quotas, safety) and soft optimizations (diversity, blending) are orchestrated in a prioritized pipeline to transform ranked candidates into a final display list that balances user experience, platform health, and operational needs.

MMRdiversityhard constraints

0 likes · 40 min read

Recommender System Re-ranking: List-Level Optimization Under Multiple Constraints

Fei's Miscellaneous Talks

Sep 6, 2026 · Artificial Intelligence

How Prediction Servers Score 1000 Candidates in Milliseconds: Fine-Ranking Architecture Deep Dive

This article details the architecture and optimization of a Prediction Server for fine-ranking in recommender systems, covering model management, batch inference, multi-objective fusion, probability calibration, deployment pipelines, and high-availability patterns to score thousands of candidates within milliseconds.

Batch InferenceModel ServingMulti-Objective Optimization

0 likes · 46 min read

How Prediction Servers Score 1000 Candidates in Milliseconds: Fine-Ranking Architecture Deep Dive

Fei's Miscellaneous Talks

Aug 31, 2026 · Artificial Intelligence

Multi-Path Recall Architecture: From Tag & Vector Search to Recall Manager Engineering

This article details the multi-path recall architecture in recommender systems, covering five recall sources (tag, vector, popularity, collaborative filtering, exploration), the Recall Manager's parallel scheduling and circuit breaking, engineering deployment pipelines with offline/online evaluation, and future trends like generative recall and joint optimization.

A/B testingANN vector searchRecall Manager

0 likes · 38 min read

Multi-Path Recall Architecture: From Tag & Vector Search to Recall Manager Engineering

Fei's Miscellaneous Talks

Jun 15, 2026 · Artificial Intelligence

Evolution of Recommendation System Architecture: From Hand‑Crafted Rules to the Large‑Model Era

This article traces the five‑generation evolution of recommendation algorithms—from manual rules, through LR and GBDT, to deep learning and large models—explaining each stage’s suitable scenarios, engineering upgrades, trade‑offs, and why practical architecture choices should prioritize business needs over chasing the newest technology.

Algorithm EvolutionGBDTMachine Learning

0 likes · 32 min read

Evolution of Recommendation System Architecture: From Hand‑Crafted Rules to the Large‑Model Era

Fei's Miscellaneous Talks

Jun 12, 2026 · Artificial Intelligence

Online Decision-Making vs. Offline Learning in Recommendation System Architecture

This article outlines the industrial‑grade offline architecture of a recommendation system, detailing how online services handle real‑time decisions while a comprehensive offline pipeline processes data, builds user profiles, extracts features, and trains models to continuously improve personalization.

Data EngineeringFeature PipelineModel Training

0 likes · 21 min read

Online Decision-Making vs. Offline Learning in Recommendation System Architecture

Fei's Miscellaneous Talks

Jun 2, 2026 · Industry Insights

Inside Industrial-Scale Recommendation Systems: From Millions of Items to a Few Real-Time Results

The article explains why a recommendation system is fundamentally an engineering platform first and an algorithm suite second, detailing its four-layer architecture, user and content modeling, multi-objective ranking, request lifecycle, and the key engineering challenges that affect scalability and real-time performance.

engineeringoffline-online collaborationreal-time processing

0 likes · 17 min read

Inside Industrial-Scale Recommendation Systems: From Millions of Items to a Few Real-Time Results