Loading…
Recommendation Systems
16 posts about Recommendation Systems. Every summary links to the original.
From Prediction to Action: How to Turn AI Outputs Into Decisions
Salesforce addressed an operational challenge where sellers faced roughly 12,000 dashboards and over 20 applications outputting machine learning predictions without clear next steps. The engineering team reframed machine learning outputs as raw signals rather than standalone answers. To bridge the gap between assessment and action, they built a Next Best Action layer that combines model signals, business logic, and contextual institutional knowledge into actionable recommendations. They integrated this layer with an AI agent using Model Context Protocol (MCP) tool contracts, enabling dynamic discovery and explicit handling of missing data. Finally, rather than introducing a separate dashboard destination, the agent serves on-demand recommendations directly inside Slack where sellers already collaborate.
Scott NybergMeta ·
MTIA 300: Meta’s First Training Chip with Built-in NICs and Communication-Offloading Engines
Training recommendation models poses distinct network bottlenecks because massive embedding tables trigger frequent collective operations that compete with computation on standard GPUs. Meta designed the MTIA 300 accelerator to overcome this by embedding two network chiplets containing twelve 800 Gbps RDMA NICs directly inside the chip package. The architecture offloads communication from its compute grid to sixteen dedicated message engines equipped with RISC-V cores and near-memory compute reduction blocks. Working alongside the co-designed HCCL library, the chip autonomously compiles and executes collective communication subgraphs without host CPU intervention. On a 150-billion-parameter production recommendation model, MTIA 300 delivered communication speeds 3.9 times faster than an equivalent GPU cluster while keeping concurrent compute degradation under 0.5 percent.
By Rajiv Krishnamurthy, Wes BlandMeta ·
From User Sequences to Scaling Laws: A Multi-Stage Architecture for Meta’s Ads Ranking
Scaling sequence models for real-time ad recommendations faces latency bottlenecks and scaling ceilings when hybrid configurations process both long user histories and sparse feature interactions online. Meta addresses these trade-offs with a multi-stage architecture that decouples heavy, asynchronous offline user modeling from lightweight, real-time online ranking. Deep upstream transformer models process sequences of thousands of actions to generate cached user embeddings, which downstream online models combine with candidate ad features using dense tokenization and target-aware multi-head attention. This framework produces predictable, LLM-style log-linear scaling curves between compute and normalized entropy without disproportionate increases in online serving costs. Implemented as part of Meta's Generative Ads Recommendation Model, the design achieved a cumulative 6% lift in Instagram conversions and a 3% lift in Facebook conversions.
By Steven De Gryze, Parshva Doshi, Sean O'Byrne, Arnold Overwijk, Dinesh Ramasamy, Lee XiongMeta ·
GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model
Meta's Generative Ads Recommendation Model (GEM) trains across thousands of GPUs using a hybrid architecture featuring trillions of sparse embedding parameters and billions of dense parameters. Standard training infrastructure built for large language models struggles with recommendation-specific workloads characterized by jagged user activity histories, asymmetric attention patterns, and numerical sensitivity. To overcome these bottlenecks, engineering teams introduced custom kernels like Jagged Flash Attention and Generalized Dot-Product Attention alongside MXFP8 mixed-precision execution. Pairing these kernel optimizations with topology-aware 5D parallelism doubled end-to-end training efficiency to 20–25% Model FLOPs Utilization while scaling total training FLOPs fourfold over twelve months.
Netflix ·
GenRec: Towards LLM-Native Recommendation at Netflix
Netflix developed GenRec, an LLM-backed recommendation ranker built by adapting an internal foundation model for large-scale personalization. Traditional recommendation stacks rely on thousands of hand-crafted features and specialized architectures that are costly to maintain across diverse content types. To replace manual feature pipelines, GenRec verbalizes user histories, metadata, and contexts into natural-language prompts and trains with multi-objective losses, including catalog-aware ranking and reward-weighted alignment. At inference time, the model executes in prefill-only mode on vLLM without decoding text. In large-scale online A/B testing against a mature production ranker, GenRec achieved statistically significant improvements in short-term and long-term metrics while using fewer labeled examples.
Netflix Technology BlogGrab ·
Crowdsourced taxonomy verification: A feedback-driven framework for refining knowledge graph relationships via online search interactions
Maintaining accurate Knowledge Graphs in dynamic domains like e-commerce and food delivery is challenging because automated language models frequently hallucinate relationships while manual curation cannot scale. To validate structural taxonomy links continuously, a closed-loop verification framework operationalizes search interfaces by injecting unverified candidate edges as hypotheses into live user traffic. The system uses an exploration-exploitation strategy to place candidate relationships in lower-risk interface slots, tracking contextually anchored micro-interactions such as clicks, dwell times, and purchases. An offline verification engine aggregates these weighted interactions into normalized confidence scores, automatically promoting verified links to permanent graph edges and pruning refuted relationships.
Junpeng NiuAirbnb ·
Personalizing Airbnb search by learning from the guest journey
Airbnb replaced hundreds of hand-crafted ranking features with a Transformer-based sequence model that captures both long-term booking history and short-term browsing behavior. To keep latency low and throughput high, sequence representations are generated via daily batch jobs and combined with real-time queries using a co-trained setwise ranker.
Daochen ZhaMeta ·
Exploring Hierarchical Interest Representation For Meta Ads Deep Funnel Optimization
Meta developed Hierarchical Interest Representation as an upstream representation layer to optimize deep funnel ad ranking across its platforms. The architecture organizes interactions among users, advertisers, and products into a heterogeneous, time-decayed graph and projects sparse raw edges into multi-level super-graphs of latent interest primitives. To encode these structures, a transformer-based hierarchical encoder incorporates multimodal product semantics processed via language models alongside topology-aware attention biases. Computation scales across billions of entities by pairing deep hash ID embeddings with FlexAttention to evaluate graph biases on the fly without materializing full matrices. Self-supervised cross-view distillation trains student representations against broader teacher graph views to yield universal embeddings and interest tokens.
By Yuhui Ouyang, Di Wang, Sreedal Menon, Jie TianNetflix ·
GenPage: Towards End-to-End Generative Homepage Construction at Netflix
Netflix traditionally constructs its structured two-dimensional homepage through a complex, multi-stage recommender pipeline that separates candidate generation and ranking across rows and entities. To simplify this architecture and optimize directly for whole-page user satisfaction, Netflix developed GenPage, an end-to-end generative transformer model that autoregressively builds the entire homepage from raw tokenized context. The system relies on a domain-specific tokenizer to compress engagement history, context injection for cold start, hybrid row decoding to minimize decoding steps, and reinforcement learning post-training. In online A/B testing against the production baseline, GenPage delivered statistically significant gains in core user engagement metrics while cutting end-to-end serving latency by 20 percent. Offline evaluations further showed that enriching context prompts improved recommendation quality more effectively than increasing model capacity in the current operating regime.
Netflix Technology BlogGrab ·
User foundation models for Grab
Traditional recommendation systems at Grab relied on siloed, manually engineered features that struggled to capture sequential interaction data and cross-service user intent across its superapp ecosystem. To resolve these limitations, Grab developed a custom user foundation model built on a transformer architecture designed to learn from both tabular profile data and time-series clickstream interactions. The architecture unifies diverse data modalities—including categorical identifiers, numerical values, text, and geographical coordinates—by framing inputs as key-value token pairs. Custom positional embeddings and attention masks ensure the model processes tabular attributes as unordered sets while maintaining the sequential order of time-series events. The pre-trained model generates holistic user embeddings and supports direct fine-tuning across downstream tasks such as fraud detection, churn prediction, and advertisement optimization.
Abhinav RaiGrab ·
Graph modelling guidelines
Graph modelling uses graph theory to represent real-world entities, relationships, and properties through nodes and edges, unlocking insights across interconnected datasets. The process begins with defining domain concepts, mapping entities and relationships, and assigning relevant descriptive properties to both nodes and edges. Practitioners choose between graph structures such as property graphs or Resource Description Framework models before optionally developing a structural schema. Data is then imported or generated to populate storage systems such as Neo4j, Amazon Neptune, Azure Cosmos DB, or in-memory data structures. Finally, the graph is visualised and analysed using algorithmic methods like shortest path calculation, centrality measurement, and community detection.
Wenxiang LuGrab ·
Scaling marketing for merchants with targeted and intelligent promos
Grab previously relied on globally assigned, heuristic promotional campaigns where all users could redeem offers until limits were reached. This lack of targeting and customisation failed to optimize promo spending or meet specific merchant business objectives. To solve this, the engineering team built Bullseye, an automated AI-driven promotional assignment system that customises and assigns offers to consumers. The architecture integrates an operations user interface, backend services, Amazon S3 storage, and Spark jobs executing a suite of predictive modules including eater segmentation, campaign impact simulation, and customer response modeling. Since deploying Bullseye in 2021, the system has increased food campaign sales while decreasing promo spend across food and Mart campaigns.
Sharon TengGrab ·
Stepping up marketing for advertisers: Scalable lookalike audience
Grab's legacy lookalike audience generation platform suffered from long creation SLAs of two working days, high costs, and low weekly update frequencies. To resolve these bottlenecks, the engineering team designed an embedding-based platform powered by an in-memory retrieval service and automated update pipelines. The system creates audience representations by averaging constituent passenger embeddings, determining user membership through real-time cosine score thresholds. To eliminate feature store latency while fitting all embeddings into memory, a hash-based compression method cuts passenger embedding storage needs by roughly 90%. Consequently, audience availability dropped to within 15 minutes of campaign creation, audience generation costs fell by 98%, and ad impressions and clicks doubled.
William WuGrab ·
Designing products and services based on Jobs to be Done
The Jobs to be Done (JTBD) framework helps organizations understand consumer goals by defining a job as the progress a consumer seeks to achieve in a specific context. To prioritize new features, GrabFood applied this framework and force diagrams to investigate consumer motivations and friction points. Interviews revealed that working parents ordering family meals experienced anxiety from navigating excessive individual choices during checkout. To address this obstacle, Grab developed an algorithm within the GrabMerchant application that identified complementary food items and automatically suggested ready-made bundles. Following the release, thousands of merchants adopted one-tap bundle creation, streamlining ordering for families while supporting restaurant growth.
Tim LangeGrab ·
Automating Multi-Armed Bandit testing during feature rollout
Traditional feature rollouts and Multi-Armed Bandit testing operate as separate workflows that often depend on delayed offline analysis. To eliminate manual intervention, the Multi-Armed Bandit Optimiser automates testing concurrently during feature rollouts by responding to minute-level feedback metrics. The architecture connects Kafka Streams data processing, a metrics server with Spark jobs, and an adaptive rollout module updating online experimentation configurations. Candidate models are evaluated via Thompson Sampling on Beta distributions, with Monte Carlo simulations determining traffic allocation across user entities. In production for the GrabFood recommendation widget, the system optimizes the Effective Conversion Rate over a 30-minute window and includes fallback distribution logic.
Weicheng ZhuGrab ·
Making Grab’s Everyday App Super
Grab manages an expanding superapp ecosystem comprising ride-hailing, food delivery, payments, and partner content surfaced through the Grab Feed. As content volume grows, the platform risks overwhelming users with irrelevant information. To address this, Grab built a recommendation engine that ranks cards using signals across user profiles, content metadata, and contextual factors such as time and location. The system employs multiple recommendation strategies—including popularity metrics, user favorites, collaborative filtering, habitual patterns, and cross-platform deep embeddings—which are selected or aggregated. Recommendation quality is evaluated via offline metrics like Recall@K and NDCG alongside online engagement experiments.
Justin Bolilia