The ML Engineer
LightNews LightNews
ml.blaze.email
The ML Engineer
@ml.blaze.email
9 followers 1 following 290 posts
Posts about ML engineering, MLOps and ML research (and a sprinkling of AI). Weekly newsletter digest at ml.blaze.email. Maintained by @alastairrushworth.com.
Posts Replies Media Videos
Pinned
ml.blaze.email
The ML Engineer @ml.blaze.email · 7h
The latest edition of The ML Engineer has dropped! Check out all the latest news and posts from the ML community in a single digest.

https://ml.blaze.email/archive/the-ml-engineer-13-01-2026/

#machinelearning #datascience
The ML Engineer 13-01-2026
🔧 Company Engineering Blogs Why We Use Separate Tech Stacks for Personalization and Experimentation (engineering​.atspotify​.com). Spotify separates ML...
ml.blaze.email
ml.blaze.email
Book Review: Transformers In Action

http://sujitpal.blogspot.com/2026/01/book-review-transformers-in-action.html

#machinelearning #datascience
Book Review: Transformers In Action
Summary: Book review covers Transformers in Action: architecture basics, LLMs, PEFT, RAG, multimodal models, SLMs, training, deployment, and ethics.
sujitpal.blogspot.com
January 13, 2026 at 12:13 PM
ml.blaze.email
The latest edition of The ML Engineer has dropped! Check out all the latest news and posts from the ML community in a single digest.

https://ml.blaze.email/archive/the-ml-engineer-13-01-2026/

#machinelearning #datascience
The ML Engineer 13-01-2026
🔧 Company Engineering Blogs Why We Use Separate Tech Stacks for Personalization and Experimentation (engineering​.atspotify​.com). Spotify separates ML...
ml.blaze.email
January 13, 2026 at 7:41 AM
ml.blaze.email
Using my new datalake in practise: a case study

https://boerman.dev/posts/amunanalytics/datalakeintro/

#machinelearning #datascience
Using my new datalake in practise: a case study
Summary: Case study of a Datalake for analytics using Python/R, Trino, OAuth2, and Iceberg on Amun Analytics
boerman.dev
January 13, 2026 at 6:13 AM
ml.blaze.email
PyTorch CUDA Graph Capture

https://leimao.github.io/blog/PyTorch-CUDA-Graph-Capture/

#machinelearning #datascience
PyTorch CUDA Graph Capture
Summary: Luaired overview of PyTorch CUDA Graphs using torch.cuda.graph and torch.cuda.make_graphed_callables in training a model with MLP, profiling, and replay
leimao.github.io
January 13, 2026 at 12:13 AM
ml.blaze.email
Why Julia’s GPU Accelerated ODE Solvers are 20x-100x Faster than Jax and PyTorch

https://www.juliabloggers.com/why-julias-gpu-accelerated-ode-solvers-are-20x-100x-faster-than-jax-and-pytorch/

#machinelearning #datascience
Why Julia’s GPU Accelerated ODE Solvers are 20x-100x Faster than Jax and PyTorch
Summary: Julia GPU-accelerated ODE solvers outperform JAX and PyTorch on GPUs, with insights into architectural differences.
www.juliabloggers.com
January 12, 2026 at 6:14 PM
ml.blaze.email
20,000 healthy GPUs

https://modal.com/blog/gpu-health

#machinelearning #datascience
20,000 healthy GPUs
Summary: Modal describes monitoring 20,000+ GPUs across AWS, GCP, Azure, and OCI with passive/active checks, DCGM, and image automation.
modal.com
January 12, 2026 at 12:13 PM
ml.blaze.email
A Tuesday in 2027: What orchestrated LLMs might look like

https://justin.abrah.ms/blog/2026-01-09-a-tuesday-in-2027-what-orchestrated-llms-might-look-like.html

#machinelearning #datascience
A Tuesday in 2027: What orchestrated LLMs might look like
Summary: Imagines orchestrated LLMs in 2027 with specs-driven, agent-enabled deployments, reduced p99 latency, and apprenticeship-style collaboration.
justin.abrah.ms
January 12, 2026 at 6:13 AM
ml.blaze.email
Using neural networks in R is still not obsolete in 2026

https://joshuamarie.com/posts/15-nn-r/

#machinelearning #datascience
Using neural networks in R is still not obsolete in 2026
Summary: Explores R's neural networks in 2026, focusing on torch, tidymodels, and the kindling package for streamlined deep learning in R.
joshuamarie.com
January 12, 2026 at 12:13 AM
ml.blaze.email
Using MLflow-RAGAS Integration Without Tracing

https://www.safjan.com/using-mlflowragas-integration-without-tracing/?utm_source=rss&utm_medium=feed&utm_campaign=safjan-blog

#machinelearning #datascience
Using MLflow-RAGAS Integration Without Tracing
Summary: Guides static-data and simple predict_fn approaches to evaluate RAG pipelines with MLflow RAGAS using Python.
www.safjan.com
January 11, 2026 at 6:12 PM
ml.blaze.email
vLLM Quickstart: High-Performance LLM Serving

https://www.glukhov.org/post/2026/01/vllm-quickstart/

#machinelearning #datascience
vLLM Quickstart: High-Performance LLM Serving
Summary: High-throughput LLM serving with vLLM: OpenAI-compatible API, PagedAttention, CUDA optimization, and multi-GPU setups.
www.glukhov.org
January 11, 2026 at 12:13 PM
ml.blaze.email
What a Platform Engineer Is (and why it’s basically DevOps with a sharper purpose)

https://chris.funderburg.me/what-is-a-platform-engineer/

#machinelearning #datascience
What a Platform Engineer Is (and why it’s basically DevOps with a sharper purpose)
Summary: Platform Engineering blends DevOps with a product mindset, expanding to data, ML, and AI infrastructure using IaC, CI/CD, Kubernetes, and observability.
chris.funderburg.me
January 11, 2026 at 6:13 AM
ml.blaze.email
Week 5 in Data Science: Image recognition neural network with 90% accuracy - a trivial task

https://igorstechnoclub.com/deep-learning-transfer-learning/

#machinelearning #datascience
Week 5 in Data Science: Image recognition neural network with 90% accuracy - a trivial task
Summary: Transfer learning with VGG16 in Google Colab to classify flower types using Keras, Python, and preprocessing.
igorstechnoclub.com
January 11, 2026 at 12:13 AM
ml.blaze.email
Quick Take: Bridging Compile-Time and Runtime Performance in Lean 4

http://alok.github.io/2026/01/07/leanbench-runtime-profiler/

#machinelearning #datascience
Quick Take: Bridging Compile-Time and Runtime Performance in Lean 4
Summary: LeanBench now correlates compilation metrics with runtime data from profilers, supporting CPU, GPU, memory, and FFI timings.
alok.github.io
January 10, 2026 at 6:13 PM
ml.blaze.email
Constrained Fuzziness: A Control System Pattern for Probabilistic Components

https://www.mostlylucid.net/blog/constrained-fuzziness-pattern

#machinelearning #datascience
Constrained Fuzziness: A Control System Pattern for Probabilistic Components
Summary: Constrained Fuzziness blends LLM proposals with deterministic constraints, budgets, and gates to ensure safe, bounded outputs in multi-domain systems.
www.mostlylucid.net
January 10, 2026 at 12:12 PM
ml.blaze.email
What Does the Linear Representation Hypothesis Even Mean?

http://alok.github.io/2026/01/08/linear-representation-hypothesis/

#machinelearning #datascience
What Does the Linear Representation Hypothesis Even Mean?
Summary: Explores what a 'linear representation' means in AI, comparing word analogies, logistic probes, and steering vectors.
alok.github.io
January 10, 2026 at 6:13 AM
ml.blaze.email
rtopy: an R to Python bridge – novelties

https://thierrymoudiki.github.io/blog/2026/01/08/r/python/rtopy

#machinelearning #datascience
rtopy: an R to Python bridge – novelties
Summary: rtopy enables translating R to Python via an enhanced RBridge and call_r, showcasing SVM with e1071, forecast ARIMA, randomForest, dplyr, and clustering examples in Python.
thierrymoudiki.github.io
January 10, 2026 at 12:13 AM
ml.blaze.email
Integración de los Wold Models de Lecun en la IAGC de Farnós (Educación disruptiva & IA)

https://juandomingofarnos.wordpress.com/2026/01/08/integracion-de-los-wold-models-de-lecun-en-la-iagc-de-farnos-educacion-disruptiva-ia/

#machinelearning #datascience
Integración de los Wold Models de Lecun en la IAGC de Farnós (Educación disruptiva & IA)
Summary: Integration of LeCun's World Models with Farnós's Low-Entropy IAGC for causal AI, FractalNet, ARC, and immutable causal hashes in education
juandomingofarnos.wordpress.com
January 9, 2026 at 6:13 PM
ml.blaze.email
RAG Evaluation with RAGAS and MLflow - A Practical Guide

https://www.safjan.com/ragas-mlflow-rag-evaluation-tutorial/?utm_source=rss&utm_medium=feed&utm_campaign=safjan-blog

#machinelearning #datascience
RAG Evaluation with RAGAS and MLflow - A Practical Guide
Summary: A practical RAG evaluation guide using RAGAS metrics with MLflow, LangChain, and OpenAI-like backends
www.safjan.com
January 9, 2026 at 12:13 PM
ml.blaze.email
Semantic Search Without Embeddings

http://softwaredoug.com/blog/2026/01/08/semantic-search-without-embeddings.html

#machinelearning #datascience
Semantic Search Without Embeddings
Summary: Explores semantic search beyond embeddings, comparing simple feature-based similarity, taxonomy-driven retrieval, and LLM-augmented categorization.
softwaredoug.com
January 9, 2026 at 6:12 AM
ml.blaze.email
Human embeddings

https://www.trfetzer.com/human-embeddings/

#machinelearning #datascience
Human embeddings
Summary: Explores human embeddings for team design using 1024D vector space, open-source embeddings, GPT-OSS, and clustering to form diverse groups.
www.trfetzer.com
January 9, 2026 at 12:12 AM
ml.blaze.email
Writing an LLM from scratch, part 29 -- using DistributedDataParallel to train a base model from scratch in the cloud

https://www.gilesthomas.com/2026/01/llm-from-scratch-29-ddp-training-a-base-model-in-the-cloud

#machinelearning #datascience
Writing an LLM from scratch, part 29 -- using DistributedDataParallel to train a base model from scratch in the cloud
Summary: DistributedDataParallel training of a base model in the cloud using PyTorch, DDP setup, and checkpointing on AWS-like GPUs
www.gilesthomas.com
January 8, 2026 at 6:12 PM
ml.blaze.email
Neural Networks: Zero to Hero

https://karpathy.ai/zero-to-hero.html

#machinelearning #datascience
Neural Networks: Zero to Hero
Summary: A practical guide to neural networks with PyTorch and Python, detailing concepts from linear models to deep learning, including backpropagation and optimization.
karpathy.ai
January 8, 2026 at 12:13 PM
ml.blaze.email
Why 90% of AI Agents Never Leave the Demo

https://mikulskibartosz.name/why-ai-agents-never-leave-demo

#machinelearning #datascience
Why 90% of AI Agents Never Leave the Demo
Summary: Practical AI engineering guide exploring three production pitfalls and how to build reliable AI pilots with abstractions, metrics, and real-data testing.
mikulskibartosz.name
January 8, 2026 at 6:13 AM
ml.blaze.email
ONNX Multi-platform benchmark

https://karnwong.me/posts/2026/01/onnx-multi-platform-benchmark/

#machinelearning #datascience
ONNX Multi-platform benchmark
Summary: ONNX multi-platform benchmark compares inference across PC, Android, and Raspberry Pi using forest regression models in Python, Rust, Web, Kotlin, and Flutter.
karnwong.me
January 8, 2026 at 12:13 AM
ml.blaze.email
Intro to GPUs For the Research Oriented

https://hackbot.dad/writing/intro-to-gpus/

#machinelearning #datascience
Intro to GPUs For the Research Oriented
Summary: Explores GPU hardware, CUDA kernels, memory bandwidth, and MFU with CUDA examples and kernel concepts in training systems using Python, Torch, and NVIDIA GPUs.
hackbot.dad
January 7, 2026 at 12:13 PM