Top all-time · #5299
Shaped
Company blog · shaped.ai/blogaidata
Shaped is the company blog at shaped.ai, writing about ai and data. Since 2022, 14 different Hacker News readers have linked it 56 times, most often “Meta's Trillion Parameter Recommendation System”. No feed was found, so read it on the site itself.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
HN commenters link Shaped's posts on ranking-system infrastructure (serving, data layer, MLOps) and on Yann LeCun's approach to autonomous AI. The comments cite or link them without judging the writing.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
The only vector database with a feedback loop, powering personalised search and recommendations. Shaped has been acquired by Whatnot.
Most-linked posts?
- Meta's Trillion Parameter Recommendation System2 people · 3 mentions · first linked 2024-06-05
- The Vector Bottleneck: Limitations of Embedding-Based Retrieval2 people · 2 mentions · first linked 2025-09-08
- The secret sauce of TikTok’s recommendations2 people · 2 mentions · first linked 2023-02-17
- Do Large Language Models (LLMs) Reason?2 people · 2 mentions · first linked 2023-02-21
- Yann LeCun: A Path Towards Autonomous Machine Intelligence2 people · 2 mentions · first linked 2022-08-26
- LLMs – A Paradigm Shift in RecSys?2 people · 2 mentions · first linked 2023-06-16
What Hacker News says?
The other direction that isn’t explicitly mentioned in this post is the variants of SASRec and Bert4Rec that are still trained on ID-Tokens but showing scaling laws much like LLMs. E.g. Meta’s approach https://arxiv.org/abs/2402.17152 (paper write up here: https://www.shaped.ai/blog/is-this-the-chatgpt-moment-for-re... )
Comment linking /blog/is-this-the-chatgpt-moment-for-recommendation-systems
on “Improving recommendation systems and search in the age of LLMs”Submitted The secret sauce of TikTok’s recommendations as a story · 46 points
HN discussionThe short answer is: we're better at recommendations and personalization and lean towards more technical teams (e.g. even with data/ML experience). They're better at traditional search and, these days, lean towards less technical teams. Longer answer is in our blog post about it: https://www.shaped.ai/blog/shaped-vs-algolia-recommend :)
Comment linking /blog/shaped-vs-algolia-recommend
on “Launch HN: Shaped (YC W22) – AI-Powered Recommendations and Search”Submitted Your browsing behavior is being modeled as a language as a story · 36 points
HN discussionIt seems like this is a substantial step in Yann LeCun's approach to AGI https://www.shaped.ai/blog/yann-lecun-a-path-towards-autonom...
Comment linking /blog/yann-lecun-a-path-towards-autonomous-machine-intelligence
on “Meta open-sources multisensory AI model that combines six types of data”Submitted Jagged Flash Attention Optimization as a story · 24 points
HN discussionSo. I think we ought to look at this issue from two perspectives. Many times I see this conflated when commenting TikTok. First, we should talk about the "tiktok rush" as one would call it, the immense algorithmic power the platform has to keep us hooked. This is real. I think anyone that's been using tiktok for a while can attest to its incredibly powerful algorithm... and this is a result of cleverly leveraging re…
Comment linking /blog/the-secret-sauce-of-tik-toks-recommendations
on “TikTok’s plan to avoid a ban gives the US Gov't unprecedented oversight power”Submitted Meta's Trillion Parameter Recommendation System as a story · 11 points
HN discussionFollow-up posts for context (same series): Part 2 – Data Layer (feature store to prevent online/offline skew; vector DB choices and pre- vs post-filtering): https://www.shaped.ai/blog/the-infrastructure-of-modern-rank... Part 3 – MLOps Backbone (training pipelines, registry, GitOps deployment, monitoring/drift/A-B): https://www.shaped.ai/blog/the-infrastructure-of-modern-rank... Happy to share more detail (autoscali…
Comment linking /blog/the-infrastructure-of-modern-ranking-systems-part-3-the-mlops-backbone---from-training-to-deployment
on “The infrastructure behind modern ranking systems (serving, data, MLOps)”Submitted Optimizing for Multiple Objectives in Search and Recommendations as a story · 10 points
HN discussionSubmitted Cosine Similarity: Not the Silver Bullet We Thought It Was as a story · 10 points
HN discussionSubmitted Takeways from Nvidia Recommender Systems Summit as a story · 10 points
HN discussionSubmitted Vector Search with OpenAI Embeddings: Lucene Is All You Need as a story · 9 points
HN discussionSubmitted Is this the ChatGPT moment for recommendation systems? as a story · 8 points
HN discussionSubmitted Twitter's Open Source Algorithm – Unveiling the code, but not the secrets as a story · 8 points
HN discussionSubmitted From Analytics to Action as a story · 8 points
HN discussionSubmitted Improving Recommender Systems with LLMs as a story · 7 points
HN discussionSubmitted LLMs – A Paradigm Shift in RecSys? as a story · 7 points
HN discussionSubmitted Calibrating Recommendations to Better Match User Interests as a story · 6 points
HN discussionSubmitted Data-Centric AI for Ranking as a story · 6 points
HN discussionSubmitted The Vector Bottleneck: Limitations of Embedding-Based Retrieval as a story · 5 points
HN discussionSubmitted Is the OpenAI collaboration what Microsoft needs to finally take on Google as a story · 5 points
HN discussionSubmitted Asking ChatGPT about itself and the future of chatbots as a story · 5 points
HN discussionSubmitted Why your feeds are getting worse over time as a story · 5 points
HN discussionSubmitted Information Retrieval vs. Recommendations as a story · 5 points
HN discussionSubmitted Embracing Embeddings: From fragmented insights to unified understanding as a story · 4 points
HN discussionSubmitted How Modern Ranking Systems Work as a story · 3 points
HN discussionSubmitted The Vector Bottleneck: Limitations of Embedding-Based Retrieval as a story · 3 points
HN discussionSubmitted Pinterest Replaced Multiple Models with One Embedding – It Outperformed Them All as a story · 3 points
HN discussionSubmitted Rag – Enhancing LLMs with External Sources for Reliable Recommendations as a story · 3 points
HN discussion
