Uncover emerging trends and practices from the world’s most innovative software professionals to help you validate your software roadmap. Celebrating 20+ years.
Joined March 2008
- Tweets12.1K
- Following673
- Followers7.4K
- Likes1.9K
A counterintuitive lesson from a RAG build: when documents are small enough and cost isn't a constraint, skipping chunking can remove significant engineering complexity.
Lan Chu, AI Tech Lead and Senior Data Scientist, explains when not to chunk, and why structured chunking is a strong first approach when you do need it.
More sessions like this are lined up for QCon London 2027, April 13-16.
Find out more about QCon London 2027 here: buff.ly/imJ1ugF
#QConLondon #RAG #AIEngineering
When Apache Fluss elects a new leader, recovery has to account for both confirmed and in-flight data.
Anton Borisov, Principal Data Architect at Fresha, breaks down the sequence: truncate to the last confirmed high water mark, restore from a RocksDB snapshot, then replay the in-flight data into the prewrite buffer.
More sessions like this are lined up for QCon London 2027, April 13-16.
Find out more about QCon London 2027 here: buff.ly/79SDTV7
#QConLondon #ApacheFluss #DistributedSystems
Building a new ML feature traditionally means stitching together Spark, BigQuery, Flink, Redis, Kubernetes, and Airflow, repeated for every new signal.
Nikhil Simha, Co-Founder & CTO at zipline.ai, explains how that growing burden contributed to data science and ML engineering becoming separate, large teams. Chronon was built to move that repeated infrastructure behind a common system.
More sessions like this are lined up for QCon London 2027, April 13-16.
Find out more about QCon London 2027 here: buff.ly/Xq4dVca
#QConLondon #FeatureStore #MLInfrastructure
Once AI is part of a critical system, model quality is only one control. Evals, inference, reliability, and autonomous workflows matter too.
The Engineering AI Systems track at QCon San Francisco connects those decisions through six production case studies.
Review the track: buff.ly/hz1btG2
Six sessions connect applied AI architecture at BlackRock, retrieval and ranking at LinkedIn, ML serving at Netflix, evals, self-improving agents, and agentic software development.
Giving every engineer a coding agent does not create shared team capability.
Read Brian Yang’s QCon AI New York session abstract: buff.ly/pT1T519
At Spotify, running a coding agent across an entire fleet means dealing with Maven, Yarn, Bazel, and custom build scripts, all with different commands.
Jo Kelly-Fenton, Engineer at Spotify, explains how Spotify built a single verify tool that routes each repository to the right build and test system, giving the agent one interface for verification regardless of the underlying toolchain.
More sessions like this are lined up for QCon London 2027, April 13-16. Find out more about QCon London 2027 here: buff.ly/0Ut0uuP
#QConLondon #CodingAgents #DeveloperExperience
Chunking for RAG has two separate context-window constraints — and the embedding model can impose the tighter one.
Lan Chu, AI Tech Lead and Senior Data Scientist at Rabobank, explains: the largest embedding models top out at around 8,000 tokens, so chunking decisions need to account for the model creating the embeddings, not only the model generating the answer.
More sessions like this are lined up for QCon London 2027, April 13-16.
Find out more about QCon London 2027 here: buff.ly/6cJwvMB
#QConLondon #RAG #EmbeddingModels
The longer a stateful Flink job runs, the more state it accumulates — eventually managing that state takes more CPU than the computation itself.
Giannis Polyzos, Principal Streaming Architect at Ververica, explains why compute-storage separation is a central design goal in Apache Fluss: separating growing state from the stream processor changes where the maintenance burden sits.
More sessions like this are lined up for QCon London 2027, April 13-16.
Find out more about QCon London 2027 here: buff.ly/ohGB1Dr
#QConLondon #ApacheFluss #StreamProcessing