recsys
(20)Model architectures, data generation, training paradigms, and unified frameworks inspired by LLMs.
16 Mar 2025  Ā·  43 min  Ā·  recsys llm teardown survey
Sending helpful & engaging pushes, filtering annoying pushes, and finding the frequency sweet spot.
24 Dec 2023  Ā·  18 min  Ā·  teardown recsys machinelearning production
9 patterns including HITL, hard mining, reframing, cascade, data flywheel, business rules layer, and more.
23 Apr 2023  Ā·  20 min  Ā·  machinelearning engineering production recsys
My three favorite papers, 17 paper summaries, and ML and non-ML lessons.
02 Oct 2022  Ā·  14 min  Ā·  recsys engineering production
Invited keynote at the Workshop for Online Recommender Systems and User Modeling (ORSUM)
23 Sep 2022  Ā·  2 min  Ā·  recsys engineering production
Industry examples, exploration strategies, warm-starting, off-policy evaluation, and more.
08 May 2022  Ā·  14 min  Ā·  teardown recsys machinelearning
Introducing randomness and/or learning from inherent randomness to mitigate position bias.
17 Apr 2022  Ā·  7 min  Ā·  recsys
Thinking about recsys as interventional vs. observational, and inverse propensity scoring.
10 Apr 2022  Ā·  8 min  Ā·  recsys eval machinelearning
Simple baselines, ideas, tech stacks, and packages to try.
03 Oct 2021  Ā·  5 min  Ā·  recsys deeplearning production survey
An overview of system design, candidate retrieval, and ranking, with industry examples.
15 Sep 2021  Ā·  1 min  Ā·  recsys machinelearning production
Focusing on long-term rewards, exploration, and frequently updated item.
05 Sep 2021  Ā·  13 min  Ā·  teardown recsys deeplearning
Why real-time RecSys? What does the system design look like in industry? How to build an MVP?
13 Jul 2021  Ā·  1 min  Ā·  recsys machinelearning production
Breaking it into offline vs. online environments, and candidate retrieval vs. ranking steps.
27 Jun 2021  Ā·  13 min  Ā·  teardown production engineering recsys š„
A whirlwind tour of bandits, embedding+MLP, sequences, graph, and user embeddings.
13 Jun 2021  Ā·  25 min  Ā·  teardown recsys machinelearning deeplearning
Why real-time? How have China & US companies built them? How to design & build an MVP?
10 Jan 2021  Ā·  21 min  Ā·  teardown machinelearning recsys production š„
Emphasis on bias, more sequential models & bandits, robust offline evaluation, and recsys in the wild.
27 Sep 2020  Ā·  16 min  Ā·  recsys deeplearning production survey
What I learned about measuring diversity, novelty, surprise, and serendipity from 10+ papers.
Comparing baselines (matrix factorization) against novel approaches using graphs & NLP.
14 Jan 2020  Ā·  2 min  Ā·  recsys machinelearning
Beating the baseline using Graph & NLP techniques on PyTorch, AUC improvement of ~21% (Part 2 of 2).
13 Jan 2020  Ā·  17 min  Ā·  recsys deeplearning nlp python š
Building a baseline recsys based on data scraped off Amazon. Warning - Lots of charts! (Part 1 of 2).
06 Jan 2020  Ā·  14 min  Ā·  recsys deeplearning python š
Join 10,300+ readers getting updates on machine learning, RecSys, LLMs, and engineering.