Viewing profile — vadman97

If there was only one alert criteria, that'd be simple. Our alerts can be configured for any data filters (eg. only matching logs with column `level='error'`); we would have to cre…

We insert from our alerts worker because we want the aggregation to happen per alert (with the aggregated data filtered by the particular alert definition). As each alert is evalua…

Forget the exact details, but we ran into issues with tuning the two setting you're describing on the Cloud instance (these settings were user level settings that seemed to get res…

Once we hit >100k inserts per second, async inserts didn't work well for us because we had limited control over the background async insert batching happening on the cluster. The b…

We do via the OpenTelemetry Elixir SDK (and the Phoenix extension) https://opentelemetry.io/docs/languages/erlang/

Materialized views can help change the ORDER BY with 0 downtime: * Create a new version of the table with the new ORDER BY. * Create a materialized view that will insert from the o…

Exploring bloom filter index merges would be an interesting addition. I do wish it were easier to profile merge performance to break down where most of the CPU time is being spent.…

3 nodes of 120 GiB and 30 vCPU each.

I see what you did there.

Our browser client would help with tracing what network requests are being made by said apps since it would capture all network requests. However, Shopify may restrict what our bro…

Hey HN, I'm Vadim, the CTO at highlight.io. We offer session replay to engineering teams and have had a lot of questions around client-side performance. We just published a benchma…

Presumably, they mean millions of logs a day when saying "millions of events a day", of which some subset corresponds to errors.

Our errors ingest is quite affordable ( https://www.highlight.io/pricing ) and let's you filter out data you are not interested in ingesting. Our logs ingest is even cheaper becaus…

Regarding model adaptation, we haven't yet explored a fine-tuned model, but it makes a lot of sense for a given class of errors. For a given code-base that is using highlight, the …

> Have you looked into finetuning the embedding model to your use case? Not yet, though this is definitely one of the next steps for us. The `gte-large` model we use is trained on …

Hi Hacker News! We’re Vadim and Chris from Highlight.io [1]. We do web app monitoring and are working on using LLMs/embeddings to add new functionality to our error monitoring prod…

Agreed, but the argument for causing damage still requires a human actor to execute GPT-4's advice. Should we be censoring LLMs just because we are offended by the output or find i…