Production-ready personalized recommender: two-tower retrieval, CatBoost ranking, optional LLM rerank. FastAPI + Qdrant + MLflow. Workshop by learnwithparam.com
Production-grade MovieLens-25M recommender: two-stage retrieval + re-ranking with a full MLOps lifecycle (Spark features, MLflow, FAISS, FastAPI on Kubernetes, A/B testing, drift detection, Prometheus/Grafana).
A distributed ML recommendation system — real-time streaming, multi-node distributed training, and fault-tolerant, scalable serving.
🤖 Explore and optimize rewards with Bandexa, a PyTorch-native library for Neural-Linear Thompson Sampling in contextual bandits.
Short-video recommendation reproduction: Two-Tower recall, Faiss TopK, time split, negative sampling, Recall@50/NDCG@50, ablation and badcases.
Production 2-Stage Recommendation Microservice (Two-Tower + CatBoost YetiRank + FAISS + FastAPI + Docker) on MovieLens-1M.
Two-tower retrieval recommender on MovieLens with FAISS HNSW serving and diversity reranking.
A recommender on 25M real MovieLens ratings: temporal split, two-tower retrieval, and the logQ correction that took it from losing to a popularity counter (0.018) to beating every baseline (0.085), served through this series' own HNSW
Movie recommender: TensorFlow two-tower retrieval + ranker, FAISS serving, offline A/B over model variants.
Production-oriented personalized news retrieval and ranking on Microsoft MIND
Hybrid recommender for MovieLens-25M combining collaborative filtering (matrix factorization) with item-side content features (genres) via a two-tower architecture, ANN-based candidate retrieval with FAISS, and a lightweight MLP re-ranker on top.
Two-tower reading recommender plus an off-policy 'opportunity lab' — retrieval, LambdaMART reranking and causal evaluation.
Two-stage short-video recommender (TikTok/Monolith-style): two-tower retrieval + a LightGBM ranker over real KuaiRec logs (12.5M interactions). Leakage-safe temporal eval; retrieval beats popularity at every k; ranking AUC 0.83; ~1ms /recommend.
Which repos would you have starred already, if you had seen them? Two-tower retrieval over public star histories, LLM fingerprints for cold start, a librarian you can talk to. Built on Hopsworks. #012 awesome-ml-systems
PyTorch two-tower (dual-encoder) retrieval recommender with in-batch negatives and Recall@K/NDCG evaluation.
PyTorch-native contextual bandits with Neural Thompson Sampling for scalable exploration and large action sets.
Active flagship: temporal ranking, two-tower retrieval, ANN trade-offs, and offline/online evaluation.
Leakage-aware two-stage recommender with Two-Tower, FAISS, TF-IDF, LambdaRank, FastAPI, Redis, Docker, and reproducible offline evaluation.
Production hybrid recommender with implicit ALS, content retrieval, diversity ranking, FastAPI, Redis, PostgreSQL and a complete MLOps stack.
To associate your repository with the two-tower topic, visit your repo's landing page and select "manage topics."