Saurav Bhattacharya

Software Engineer with 12+ yrs in Big Tech. I write about agent reliability, rigorous evaluation frameworks & practical AI safety to make autonomous systems reliable at scale.

Seattle WA Joined on Apr 5, 2026 https://github.com/sauravbhattacharya001

Saurav Bhattacharya

Aug 12

Measure the Judge Before You Trust It: Self-Consistency Comes Before Human Agreement

#ai #evaluation #dotnet #testing

6 min read

Want to connect with Saurav Bhattacharya?

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

Create Account

Already have an account? Sign in

Saurav Bhattacharya

Aug 12

Your Agent's Context Window Overflowed and It Answered Anyway

#ai #agents #observability #evaluation

4 min read

Saurav Bhattacharya

Aug 9

Your Golden Dataset Is Rotting: The Eval Oracle Nobody Re-Validates

#ai #agents #evaluation #observability

5 min read

Saurav Bhattacharya

Aug 7

Your Agent Passes Every Turn and Fails the Conversation

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Aug 5

Your Agent Said It Worked. Go Check the World, Not the Sentence.

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Aug 2

Your New Eval Rule Is Untested Code Guarding Production

#ai #agents #evaluation #testing

5 min read

Saurav Bhattacharya

Jul 31

Your Agent's Deadline Is a Correctness Test, Not an SLO

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Jul 29

Your Agent's Confidence Score Is Not a Probability

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Jul 26

Right Tool, Wrong Arguments: The Agent Failure Your Evals Wave Through

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Jul 24

Stop Leaving Findings in the Judge: The Ratchet That Turns Opinions Into Gates

#ai #agents #evaluation #observability

4 min read

Saurav Bhattacharya

Jul 22

The Cold-Start Problem for Agent Evals: What to Gate on Day One With Zero Labeled Data

#ai #agents #evaluation #typescript

4 min read

Saurav Bhattacharya

Jul 19

Stop Judging Every Run: Eval Sampling Is a Budget Decision, Not a Coverage One

#ai #agents #evaluation #observability

5 min read

Saurav Bhattacharya

Jul 15

Prompt Injection Is a Tier 1 Problem: Stop Asking a Judge to Guard Your Agent's Boundary

#ai #agents #safety #observability

5 min read

Saurav Bhattacharya

Jul 8

A Failed Eval Is a Decision: What Your Agent Should Actually Do When a Gate Goes Red

#ai #agents #safety #observability

5 min read

Saurav Bhattacharya

Jul 2

One Triage Pass, Every Trace Format: Stop Letting Fragmentation Shrink Your Eval Coverage

#ai #agents #observability #evaluation

5 min read

Saurav Bhattacharya

Jul 2

Short-Circuit Your Agent Evals: Tier Order Is a Latency Budget, Not a Preference

#ai #agents #evaluation #typescript

5 min read

Saurav Bhattacharya

Jul 1

Your Agent's Retries Are Double-Charging Your Users (and Every Eval Is Green)

#ai #agents #observability #typescript

5 min read

Saurav Bhattacharya

Jun 30

Your Model Upgrade Broke Three Workflows and the Tests Still Passed

#ai #agents #evaluation #testing

5 min read

Saurav Bhattacharya

Jun 29

Governing What AI Can Execute: From Product Compliance to Sovereign Gatekeeping

#ai #security #governance #cybersecurity

4 min read

Saurav Bhattacharya

Jun 29

Give Your Agent a Type Signature: Contract-First Output Beats a Smarter Judge

#ai #agents #evaluation #typescript

4 min read

Saurav Bhattacharya

Jun 28

Your Model-as-Judge Doesn't Belong in the Hot Path

#ai #agents #evaluation #observability

9 min read

Saurav Bhattacharya

Jun 27

Who Grades the Grader? Your LLM Judge Is an Unvalidated Model in Production

#ai #evaluation #observability #testing

5 min read

Saurav Bhattacharya

Jun 26

Your Agents Are Fine. The Handoff Between Them Isn't.

#ai #agents #evaluation #observability

5 min read

Saurav Bhattacharya

Jun 25

Your Evals Are Flaky Too: Stop Trusting a Pass Rate You Can't Reproduce

#ai #evaluation #testing #typescript

5 min read

Saurav Bhattacharya

Jun 24

You Can't Reproduce Your Agent's Bugs—That's Why You Can't Fix Them

#ai #agents #testing #observability

6 min read

Saurav Bhattacharya

Jun 22

Shadow Deployments for AI Agents: Canary Your Prompt Changes Before They Burn Production

#ai #agents #evaluation #observability

5 min read

Saurav Bhattacharya

Jun 22

Harness Engineering Has No Fixed Address

#ai #api #architecture

7 min read

Saurav Bhattacharya

Jun 21

The Agent Is the Harness, Not the Model — and Why That Reorganizes Software Engineering

#ai #agents #softwareengineering #career

5 min read

Saurav Bhattacharya

Jun 21

Goodhart's Law Comes for Your Agent Evals: Why Your Green Dashboard Stops Meaning Anything

#ai #agents #evaluation #observability

5 min read

Saurav Bhattacharya

Jun 20

Agent = Model x Harness: Your Eval Layer Is Part of the Agent, Not a Tool Beside It

#ai #agents #evaluation #observability

6 min read

Saurav Bhattacharya

Jun 20

Your Agent Didn't Break, It Drifted: Detecting Slow Decay in Autonomous Systems

#ai #agents #observability #evaluation

7 min read

Saurav Bhattacharya

Jun 19

Hallucination Is Not a Vibe: How to Actually Detect Ungrounded Claims in Agent Output

#ai #agents #evaluation #observability

6 min read

Saurav Bhattacharya

Jun 18

Your Agent Passed Every Eval and Still Cost $4,000 a Day

#ai #agents #observability #testing

5 min read

Saurav Bhattacharya

Jun 17

Your Eval Suite Is Grading Fiction: Stop Inventing Test Cases and Mine Your Traces

#ai #evaluation #observability #typescript

5 min read

Saurav Bhattacharya

Jun 16

Put Your Agent Evals in CI or Stop Calling Them Evals

#ai #agents #evaluation #devops

5 min read

Treats agents as distributed systems

Saurav Bhattacharya

Jun 14

The Five Agent Failure Modes Nobody Catches in Staging

#ai #agents #observability #testing

5 min read

Saurav Bhattacharya

Jun 13

Your Agent Logs Are Lying to You: What to Actually Trace in an Agentic System

#ai #agents #observability #typescript

5 min read

Saurav Bhattacharya

Jun 12

Stop Asserting Equality: How to Test Agents When Every Run Is Different

#testing #ai #agents #typescript

5 min read

Saurav Bhattacharya

Jun 11

The Reason Your Agent Demo Isn't in Production Has Nothing to Do With the Model

#agents #ai #observability #testing

4 min read

Saurav Bhattacharya

Jun 9

I Tested Claude Opus 4, GPT-4.1, GPT-4o, Sonnet 4, and Gemini 2.5 Pro on 10 Adversarial Scenarios. They All Broke on the Same One.

#ai #agents #safety #evaluation

11 min read

Saurav Bhattacharya

Jun 8

I Built an Adversarial Eval Framework and Attacked 5 LLMs — Every Single One Failed

#ai #llm #security #testing

8 min read

Saurav Bhattacharya

Jun 8

Hallucination Detection Is Not a Model Problem—It's an Infrastructure Problem

#ai #observability #testing #typescript

4 min read

Saurav Bhattacharya

Jun 7

The Alignment Problem Is an HR Problem - And We Should Treat It Like One

#ai #agents #safety #evaluation

4 min read

Saurav Bhattacharya

Jun 7

Evals Are Alignment Enforcement: Why Your Safety Strategy Needs Runtime Checks

#ai #security #evaluation #agents

5 min read

Saurav Bhattacharya

Jun 6

Your AI Agent Drifted Last Night and You Didn't Notice

#ai #agents #testing #devops

5 min read

Saurav Bhattacharya

Jun 5

Deterministic Checks vs Model-as-Judge: A Tiered Approach to Agent Evaluation

#ai #testing #agents #evaluation

4 min read

1 Week Community Wellness Streak

Writing Debut

Want to connect with Saurav Bhattacharya?