Omer Hochman

Omer Hochman

Aug 31

You added a second SQL engine. Your text-to-SQL model is still being told it's the first one.

#sql #llm #database

4 min read

Omer Hochman

Aug 30

You added ClickHouse. Your Postgres SQL validator now rejects valid queries — quietly.

#sql #clickhouse #security

3 min read

Omer Hochman

Aug 29

We read the agent-memory benchmarks. Almost none measure analysis.

#ai #machinelearning #database

3 min read

Omer Hochman

Aug 27

We shipped 18 SEO pages and got 1 referral. The links only pointed one way.

#seo #webdev #contentmarketing

3 min read

Omer Hochman

Aug 26

The timeout that looked like a hallucination

#llm #benchmarking #eval

3 min read

Omer Hochman

Aug 25

Your "best model" toggle quietly serves the cheap model. Ship a 409 instead.

#llm #api #ux

3 min read

Omer Hochman

Aug 22

Your LLM health probe passed. Your agent still starved.

#llm #ci #testing

3 min read

Omer Hochman

Aug 19

Your BI tool got acquired. Your data layer shouldn't have to care.

#dataengineering #businessintelligence #sql

3 min read

Omer Hochman

Aug 18

The duplicate-rows query you re-Google every six weeks

#sql #database #analytics

2 min read

Omer Hochman

Aug 12

Your text-to-SQL eval is lying: the gateway returns HTTP 200 with the error in the body

#llm #api #debugging

1

3 min read

Omer Hochman

Aug 11

The text-to-SQL demo takes an afternoon. The other 90% is why you should buy it.

#sql #database #ai #webdev

1

3 min read

Omer Hochman

Aug 8

Your offline LLM eval isn't measuring your model — it's measuring your rate limits

#llm #ai #testing

1

3 min read

Omer Hochman

Aug 7

Your text-to-SQL model isn't as wrong as your benchmark says. The gold SQL is.

#sql #llm #ai

1

3 min read

Omer Hochman

Aug 6

Your text-to-SQL accuracy is measured on schemas your users will never build

#sql #llm #ai

1

3 min read

Omer Hochman

Aug 3

Every data tool shipped an MCP server this year. Your agent still can't build on most of them.

#mcp #ai #api

1

2 min read

Omer Hochman

Aug 2

Your agent's memory is a vector store. Ask it "how many" and watch it fall over.

#ai #database #postgres #sql

1

2 min read

Omer Hochman

Jul 29

Your database scales to zero. Your retry loop doesn't know that.

#database #serverless #webdev

1

4 min read

Omer Hochman

Jul 28

AI made the internal-tool builder faster. It didn't ask whether you needed the tool.

#ai #database #webdev

2

3

2 min read

Omer Hochman

Jul 27

Top N per group is the query `LIMIT` can't write

#sql #database #webdev

3 min read

Omer Hochman

Jul 26

We published 20 blog posts and never shipped a feed. Nothing could subscribe.

#webdev #blogging #seo

1

4 min read

Omer Hochman

Jul 24

You don't need a backend to store form submissions. You need a place to ask "how many."

#sql #webdev #database #tutorial

1

2 min read

Omer Hochman

Jul 21

Your sitemap is advertising redirects — and your canonical tag points at one

#seo #webdev #sitemap

1

2 min read

Omer Hochman

Jul 20

NOT IN returned zero rows. It wasn't your data — it was one NULL.

#sql #postgres #database

1

2 min read

Omer Hochman

Jul 19

Zep gives my agent perfect recall. It still can't answer "average per group" about its own memory.

#ai #agents #database

1

1 min read

Omer Hochman

Jul 18

The NULL timestamp that broke a TTL sweep and a funnel metric at the same time

#database #metrics #testing

2 min read

Omer Hochman

Jul 16

The redesign shipped. The smoke test kept walking the old UI.

#testing #e2e #frontend

1

2 min read

Omer Hochman

Jul 7

Your LLM fused the two columns you asked for — and the eval marked it wrong

#sql #llm #ai

3 min read

DEV Community

Badges

Writing Debut

You added a second SQL engine. Your text-to-SQL model is still being told it's the first one.

You added ClickHouse. Your Postgres SQL validator now rejects valid queries — quietly.

We read the agent-memory benchmarks. Almost none measure analysis.

We shipped 18 SEO pages and got 1 referral. The links only pointed one way.

The timeout that looked like a hallucination

Your "best model" toggle quietly serves the cheap model. Ship a 409 instead.

Your LLM health probe passed. Your agent still starved.

Your BI tool got acquired. Your data layer shouldn't have to care.

The duplicate-rows query you re-Google every six weeks

Your text-to-SQL eval is lying: the gateway returns HTTP 200 with the error in the body

The text-to-SQL demo takes an afternoon. The other 90% is why you should buy it.

Your offline LLM eval isn't measuring your model — it's measuring your rate limits

Your text-to-SQL model isn't as wrong as your benchmark says. The gold SQL is.

Your text-to-SQL accuracy is measured on schemas your users will never build

Every data tool shipped an MCP server this year. Your agent still can't build on most of them.

Your agent's memory is a vector store. Ask it "how many" and watch it fall over.

Your database scales to zero. Your retry loop doesn't know that.

AI made the internal-tool builder faster. It didn't ask whether you needed the tool.

Top N per group is the query `LIMIT` can't write

We published 20 blog posts and never shipped a feed. Nothing could subscribe.

You don't need a backend to store form submissions. You need a place to ask "how many."

Your sitemap is advertising redirects — and your canonical tag points at one

NOT IN returned zero rows. It wasn't your data — it was one NULL.

Zep gives my agent perfect recall. It still can't answer "average per group" about its own memory.

The NULL timestamp that broke a TTL sweep and a funnel metric at the same time

The redesign shipped. The smoke test kept walking the old UI.

Your LLM fused the two columns you asked for — and the eval marked it wrong