DEV Community

Debashish Ghosal profile picture

Debashish Ghosal

TLDR - Engineer -> Manager -> Tech Leadership, passionate about AI, Developer experience, planet scale infra, transformation - AI-native leadership. Rest here - https://www.linkedin.com/in/deghosal/

Location Seattle, WA Joined Joined on  Personal website https://github.com/deghosal-2026/ github website

Pronouns

he/him

Work

Engineering Leadership

Pin Pinned

I Stopped Trusting AI Agents With Tools. So I Built a Gatekeeper.

83 real agents tested zero mocks
debashish_ghosal profile

I Stopped Trusting AI Agents With Tools. So I Built a Gatekeeper.

#ai #agents #security #gatekeeper
50
Comments 55
14 min read
I Built an Agent Eval Harness. Real Agents Broke the Clean Version of the Story

Real-world agents break clean evals
debashish_ghosal profile

I Built an Agent Eval Harness. Real Agents Broke the Clean Version of the Story

#ai #agentskills #learning #opensource
29
Comments 41
9 min read
I Added a Fourth Model Mid-Run. It Changed What My Field Test Could Prove.
debashish_ghosal profile

I Added a Fourth Model Mid-Run. It Changed What My Field Test Could Prove.

#healthydebate #ai #llm #testing
6
Comments
7 min read

Want to connect with Debashish Ghosal?

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

Create Account
Already have an account? Sign in
The Same Model Debating Itself Was More Self-Critical Than Two Different Models
debashish_ghosal profile

The Same Model Debating Itself Was More Self-Critical Than Two Different Models

#healthydebate #ai #llm #testing
13
Comments
13 min read
Two Projects, One Problem — What PlannerCritic and AdversarialDebate Each Got Wrong
debashish_ghosal profile

Two Projects, One Problem — What PlannerCritic and AdversarialDebate Each Got Wrong

#healthydebate #ai #dogfood #testing
16
Comments 2
10 min read
The Best Model Pair in My Field Test Was Also the Least Trustworthy
debashish_ghosal profile

The Best Model Pair in My Field Test Was Also the Least Trustworthy

#healthydebate #ai #trustworthy #llm
21
Comments 7
12 min read
I Thought My Multi-Agent Debate Engine Was Broken. The Real Bug Was the Prompt.
debashish_ghosal profile

I Thought My Multi-Agent Debate Engine Was Broken. The Real Bug Was the Prompt.

#ai #promptengineering #agents #testing
14
Comments
31 min read
Most AI Second Opinions Are Theater. I Built a System That Actually Fights Back.

Reveals why blind reviews fail
debashish_ghosal profile

Most AI Second Opinions Are Theater. I Built a System That Actually Fights Back.

#healthydebate #ai #testing #llm
13
Comments 8
14 min read
My LLM Critic Disagreed With Itself on Every Trial. The Safe Part Was the Code I Didn’t Trust It to Touch.
debashish_ghosal profile

My LLM Critic Disagreed With Itself on Every Trial. The Safe Part Was the Code I Didn’t Trust It to Touch.

#healthydebate #ai #testing #llm
21
Comments 5
5 min read
My Agent Refused 96 Times. That Was the Right Output.
debashish_ghosal profile

My Agent Refused 96 Times. That Was the Right Output.

#ai #planner #testing #llm
22
Comments 5
10 min read
Most AI Second Opinions Are Fake. I Built a Two-LLM Review Engine to Prove It.

Tested on 70 real PRs for just $0.53 total cost
debashish_ghosal profile

Most AI Second Opinions Are Fake. I Built a Two-LLM Review Engine to Prove It.

#ai #adversarial #llm #testing
16
Comments 12
11 min read
A Reader Audited My OSS Release in Public. He Found the Contradictions I Missed.
debashish_ghosal profile

A Reader Audited My OSS Release in Public. He Found the Contradictions I Missed.

#ai #planner #llm #testing
18
Comments 8
7 min read
I Tried to Prompt-Inject My Own Agent Engine. It Didn't Work. Here's Why.

Structural gates beat prompt safety
debashish_ghosal profile

I Tried to Prompt-Inject My Own Agent Engine. It Didn't Work. Here's Why.

#ai #agents #security #llm
40
Comments 9
12 min read
I Ran 170 Agent Goals for $0.49. The Field Test Found 10 Issues That Unit Tests Never Would.
debashish_ghosal profile

I Ran 170 Agent Goals for $0.49. The Field Test Found 10 Issues That Unit Tests Never Would.

#ai #llm #infrastructure #testing
17
Comments 5
13 min read
The Planner Made the Same 3 Mistakes Every Time. A Bigger Model Didn't Fix It.

Exposes the three recurring failure patterns
debashish_ghosal profile

The Planner Made the Same 3 Mistakes Every Time. A Bigger Model Didn't Fix It.

#ai #agents #llm #testing
21
Comments 28
5 min read
I Told My LLM Critic to Be Adversarial. It Started Blocking Plans for Being 'Not Thorough Enough.'

Code-enforced allowlists fix prompt drift
debashish_ghosal profile

I Told My LLM Critic to Be Adversarial. It Started Blocking Plans for Being 'Not Thorough Enough.'

#ai #adversarial #llm #promptengineering
16
Comments 26
4 min read
I Ran 157 Agent Plans Against a Real LLM. The Problem Wasn't Execution. It Was Planning.

Explores why self-review fails agents
debashish_ghosal profile

I Ran 157 Agent Plans Against a Real LLM. The Problem Wasn't Execution. It Was Planning.

#ai #agentskills #fieldtest #development
33
Comments 32
8 min read
AI Governance Is Becoming a Transformation Problem
debashish_ghosal profile

AI Governance Is Becoming a Transformation Problem

#ai #governance #agile
22
Comments 4
8 min read
Your Company Has AI Tribes. Send an Engineer as Emissary

Adapting Palantir's model internally
debashish_ghosal profile

Your Company Has AI Tribes. Send an Engineer as Emissary

#ai #fde #culture
7
Comments 4
12 min read
I Shipped an Agent Gatekeeper (v0.1). 14 Developers Showed Me What I Missed. Here's v0.2 — a Control Plane.
debashish_ghosal profile

I Shipped an Agent Gatekeeper (v0.1). 14 Developers Showed Me What I Missed. Here's v0.2 — a Control Plane.

#ai #agents #gatekeeper #llm
8
Comments 2
13 min read
AI Made Prototyping Free. That Is Exactly Why Your Portfolio Strategy Matters Now.
debashish_ghosal profile

AI Made Prototyping Free. That Is Exactly Why Your Portfolio Strategy Matters Now.

#ai #leadership #prioritization
13
Comments 3
13 min read
I Built Scenario Packs for Agent Regression Testing. The Integration, Not the Judge, Broke Me.

Adapters need conformance suites too
debashish_ghosal profile

I Built Scenario Packs for Agent Regression Testing. The Integration, Not the Judge, Broke Me.

#discuss #ai #agentskills
18
Comments 21
15 min read
I Thought Building Agent Observability Was a Detector Problem. I Was Wrong.

Real production data broke 35 detectors instantly
debashish_ghosal profile

I Thought Building Agent Observability Was a Detector Problem. I Was Wrong.

#ai #opentelemetry #llmops #opensource
22
Comments 19
8 min read
Why Agent Evaluation Is Harder Than Model Evaluation

Builder scars reveal untrustworthy agent paths
debashish_ghosal profile

Why Agent Evaluation Is Harder Than Model Evaluation

#challenge #ai #agents #programming
19
Comments 25
8 min read
Faster PRs, Weaker Instincts: The Judgment Problem in AI-Assisted Engineering

Metrics mask silent logic gaps
debashish_ghosal profile

Faster PRs, Weaker Instincts: The Judgment Problem in AI-Assisted Engineering

#ai #leadership #llm #agile
8
Comments 3
4 min read
AI-Assisted Engineering: Faster to Build Isn't Cheaper to Own

QA and exception handling get denser
debashish_ghosal profile

AI-Assisted Engineering: Faster to Build Isn't Cheaper to Own

#ai #leadership #llm #agentskills
13
Comments 8
6 min read
I Planned 10 LLM Evaluation Experiments And Only Ran 1. It Was Enough.
debashish_ghosal profile

I Planned 10 LLM Evaluation Experiments And Only Ran 1. It Was Enough.

#discuss #ai #webdev #llm
5
Comments 2
13 min read
I Connected 3 MCP Servers to One Agent. It Got Scary Fast.

Exposes glaring security gaps in multi-MCP setups
debashish_ghosal profile

I Connected 3 MCP Servers to One Agent. It Got Scary Fast.

#discuss #ai #mcp #automation
15
Comments 33
5 min read
Platform Strategy That Ships: The 90-Day System I Trust More Than AI Roadmaps
debashish_ghosal profile

Platform Strategy That Ships: The 90-Day System I Trust More Than AI Roadmaps

5
Comments
7 min read
4 Open-Source AI Tools, 1 MCP Server — What I Built and What I Learned

Overcoming internal dev tool adoption hurdles
debashish_ghosal profile

4 Open-Source AI Tools, 1 MCP Server — What I Built and What I Learned

#discuss #ai #python #opensource
10
Comments 17
15 min read
The Best AI Productivity Win I Got This Month Wasn't From Coding
debashish_ghosal profile

The Best AI Productivity Win I Got This Month Wasn't From Coding

#discuss #ai #productivity #softwareengineering
Comments
4 min read
Your Flaky Tests Aren't Flaky
debashish_ghosal profile

Your Flaky Tests Aren't Flaky

#discuss #ai #python #testing
Comments
6 min read
i built a tool that tracks what AI tasks actually cost. the real number surprised me.
debashish_ghosal profile

i built a tool that tracks what AI tasks actually cost. the real number surprised me.

#discuss #ai #llm
Comments 3
6 min read
i've been building platforms first for 25 years. i think it's wrong now.
debashish_ghosal profile

i've been building platforms first for 25 years. i think it's wrong now.

#discuss #ai #webdev #softwareengineering
1
Comments
6 min read
i tested an ai incident commander against 15 real outages — 88% pass rate
debashish_ghosal profile

i tested an ai incident commander against 15 real outages — 88% pass rate

#discuss #ai #webdev #python
1
Comments 2
4 min read
i spent 3 days on a rag pipeline and the embedding model didn't matter at all
debashish_ghosal profile

i spent 3 days on a rag pipeline and the embedding model didn't matter at all

#discuss #ai #webdev #rag
Comments
6 min read
my ai agent ran for 6 hours on a 2-minute task and cost me $200
debashish_ghosal profile

my ai agent ran for 6 hours on a 2-minute task and cost me $200

#discuss #ai #webdev #opensource
1
Comments 1
6 min read
i've been doing sdlc for 25 years. ai didn't kill it. it made it the coordination layer.
debashish_ghosal profile

i've been doing sdlc for 25 years. ai didn't kill it. it made it the coordination layer.

#discuss #ai #webdev #sdlc
1
Comments 1
7 min read
i was using a frontier model to update jira tickets. here's how i fixed it.
debashish_ghosal profile

i was using a frontier model to update jira tickets. here's how i fixed it.

#discuss #ai #webdev #llm
2
Comments
6 min read
i built two langgraph projects and learned that agents are junior engineers with amnesia
debashish_ghosal profile

i built two langgraph projects and learned that agents are junior engineers with amnesia

#discuss #ai #webdev #langgraph
2
Comments 8
7 min read
loading...