Vignesh Reddy

Unborn , Unthinkable , Undoable

Joined on May 30, 2026

Vignesh Reddy

Jun 14

How to add full observability to your LangChain and LlamaIndex agents in under 10 minutes

#agents #llm #monitoring #tutorial

2 min read

Want to connect with Vignesh Reddy?

Create an account to connect with Vignesh Reddy. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

Vignesh Reddy

Jun 13

Rate limiting, email alerts, health checks, and Grafana — what we shipped to make Ajah production-ready

#api #devops #llm #monitoring

2 min read

Vignesh Reddy

Jun 6

How I built narrative drift detection for LLM agent runs

#showdev #agents #llm #monitoring

1 min read

Vignesh Reddy

Jun 5

How I added real-time Slack alerts to an open-source LLM gateway in one day

#showdev #llm #monitoring #opensource

1 min read

Vignesh Reddy

Jun 4

The LLM failure mode nobody is monitoring: overconfident responses in high-stakes domains

#showdev #ai #llm #monitoring

1 min read

Vignesh Reddy

Jun 2

Helicone got acquired. Langfuse got acquired. Here's what I built instead.

1 min read

Vignesh Reddy

Jun 1

I found my own tool making twice the API calls it should. Here's what I fixed.

#api #architecture #go #performance

2 min read

Vignesh Reddy

May 31

I got tired of LLM observability tools getting acquired. So I built one that can't be.

#showdev #ai #llm #opensource

1 min read

Vignesh Reddy

May 30

Why I built Ajah after Helicone went into maintenance mode

#showdev #ai #llm #privacy

1 min read

DEV Community

Vignesh Reddy

Badges

Writing Debut

How to add full observability to your LangChain and LlamaIndex agents in under 10 minutes

Want to connect with Vignesh Reddy?

Rate limiting, email alerts, health checks, and Grafana — what we shipped to make Ajah production-ready

How I built narrative drift detection for LLM agent runs

How I added real-time Slack alerts to an open-source LLM gateway in one day

The LLM failure mode nobody is monitoring: overconfident responses in high-stakes domains

Helicone got acquired. Langfuse got acquired. Here's what I built instead.

I found my own tool making twice the API calls it should. Here's what I fixed.

I got tired of LLM observability tools getting acquired. So I built one that can't be.

Why I built Ajah after Helicone went into maintenance mode