Ganesh H N

QA Professional | 13+ yrs in Software Quality | Exploring AI & building products that solve real problems | Building AI Quality Workforce Intelligence | Learnin

[Ganesh H N @ganesh2412]

I have been building AI Quality Workforce Intelligence - a V1 demo exploring how QA teams can bring skills, capacity, workload, testing progress, and release visibility together in one workspace.

The demo includes five perspectives:

1. Head of QA: Portfolio metrics, team performance, skills coverage, capacity, and release health.
2. QA Manager: Team workload, availability, allocations, skill gaps, and project health.
3. Test Lead: Daily assignments, execution progress, blockers, and release priorities.
4. Manual QA: Assigned test work, test results, evidence, defects, and retesting.
5. Automation QA: Automation work, test runs, failures, coverage, and framework health.

The goal is to help teams answer practical questions: Who is available? Who has the right skills for this work? Where is the team overloaded? What needs attention before a release?

This is an early demo using synthetic data, not a production service. I would love feedback from QA leaders, test managers, and quality engineers. What would you want to see to make better staffing and release decisions?

Try the demo: https://t.co/QtqALerdcj

Project: https://t.co/gFHDBEyCZy

#QualityEngineering #SoftwareTesting #TestAutomation #QualityAssurance #AITesting #QualityControl

[Ganesh H N @ganesh2412]

That's a great addition, Gary! Tracking reverts within 7 days would help catch regressions that slipped through review. This could make AI coding benchmarks much more meaningful.

Happy to connect, exchange ideas, share real-world QA challenges and help each other improve AI development and testing!

[Ganesh H N @ganesh2412]

The Claude Code vs Codex debate is interesting.

But I care less about which model writes code faster and more about which one needs fewer corrections before production.

What are you actually using every day?

#ClaudeCode #Codex

[Ganesh H N @ganesh2412]

Hey Yajat! Great to connect! 👋

BUK!NN looks interesting! Managing venue bookings, availability, catering and teams in one place could solve a lot of headaches for hospitality businesses.

Coming from 11+ years in QA, I'm curious how you handle double bookings when two customers try to reserve the same venue or date simultaneously.

Are you already working with resorts and banquet owners, or still looking for early users?

[Ganesh H N @ganesh2412]

I would love to connect with developers and QA folks building with AI.

Especially those sharing failed attempts, useful fixes and things they still verify manually.

I bring 11+ years in QA and plenty of beginner-builder questions.

What are you building?

#AICoding #BuildInPublic

[Ganesh H N @ganesh2412]

@MalikAbdul94020 Hey Malik! Thanks for the follow, great to connect! 😄

I'm exploring AI, automation and building my own SaaS products after 11+ years in QA.

What are you working on these days? Would love to hear about your journey!

[Ganesh H N @ganesh2412]

Hey Daniel! Great to connect! 👋

Nice to see another founder building something of their own. I'm exploring AI and automation while working on a couple of SaaS products after 11+ years in QA.

Curious to learn more about Bevizio! What problem are you trying to solve, and what's been your biggest challenge building it so far?

Happy to exchange ideas and feedback!

[Ganesh H N @ganesh2412]

Hey Maor! Great to connect! 😄 BOOX sounds interesting, especially the part where AI generates questions but store owners can review and edit the answers.

Coming from 11+ years in QA, I really like keeping humans in control of AI-generated content.

Curious, how does BOOX handle incorrect product information or questions the AI can't confidently answer?

Happy to exchange ideas and feedback!

[Ganesh H N @ganesh2412]

Exactly! 😄 The goal isn't to send more reminders, it's to know when NOT to send one.

I'm thinking of letting users set follow-up limits, detect when a client responds, and require human approval for sensitive situations.

One wrong follow-up could damage a client relationship. Would you prefer full manual approval or automation with clear limits?

[Ganesh H N @ganesh2412]

That's exactly the danger! If AI writes both the feature and tests based on the same wrong assumption, everything can pass while the actual requirement is missed.

I would rather have independent tests based on the original requirements. A green test suite should prove the feature works, not just that the AI agrees with itself! 😄

[Ganesh H N @ganesh2412]

Haha, exactly Javi! 😂 AI can build a beautiful UI in minutes, but one real user interaction can expose what 50 passing tests missed.

That's why I think manual exploratory testing is becoming even more valuable with AI coding.

What's the most surprising bug you've caught just by clicking through an AI-built feature?

[Ganesh H N @ganesh2412]

Hey Ryan! Great to connect! 👋

Just noticed you're building SeaOtter with isolated Hermes agents and MCP. That's interesting, especially from a QA and security perspective!

After 11+ years in QA, I'm now exploring AI agents and building my own SaaS products.

Curious, how do you test agent isolation and make sure one agent can't access another's data or resources?

[Ganesh H N @ganesh2412]

Hey Amol! Great to connect with a fellow SaaS builder! 😄 I'm coming from 11+ years in QA and now building a few AI-powered SaaS products myself.

Still learning a lot along the way. Curious, what SaaS product are you building, and what's been your biggest challenge so far finding users, building features, or making the AI reliable?

[Ganesh H N @ganesh2412]

@im_kshitija You're welcome! 😊

[Ganesh H N @ganesh2412]

That's a great metric, Gary! Corrections per merged PR tells us much more than generation speed. I would also track regression bugs and unrelated changes introduced. A PR that merges quickly but breaks something later isn't really a win. Have you started measuring this across different agents?

[Ganesh H N @ganesh2412]

@muhammadmaazx Great to connect, Maaz! 😄 Always happy to meet fellow developers and builders. I'm exploring AI, automation and building my own SaaS products after 11+ years in QA.

[Ganesh H N @ganesh2412]

And that's completely okay, Kshitija! 😄 You have plenty of time. Explore different things, experiment with AI, discover what genuinely interests you and then go for it.

Don't rush to figure everything out. Even experienced people are still learning in this AI world!

Start small, stay curious and enjoy the journey. And if you ever need guidance, feel free to reach out. Happy to help!

[Ganesh H N @ganesh2412]

@Rakshaweb3 Absolutely, Raksha! Saving 10 minutes generating code means nothing if we spend 2 hours fixing it. I'd rather have an AI agent that's slower but gets things right with fewer corrections. Are you using Claude Code, Codex or something else for your daily work?

[Ganesh H N @ganesh2412]

Hey Kshitija! 😄 You don't need years of experience to start building, especially in today's AI world!

AI has made learning and experimenting more accessible than ever. Stay curious, keep learning, build small things and don't be afraid to fail.

Even after 11+ years in QA, I'm learning something new every day!

Happy to connect! If you ever need help or guidance, feel free to ask. If I know the answer, I'll gladly help. Otherwise, we can figure it out together!

What are you interested in learning or building with AI?

[Ganesh H N @ganesh2412]

@AskZaratech Thanks, Zara! 😄 I believe trust comes from what happens after the demo, especially when things go wrong. A working feature is great, but reliable error handling, security and recovery are what keep users coming back. What's your biggest concern with AI-built apps?

[Ganesh H N @ganesh2412]

Vibe coding is getting ridiculously good.

But there's a difference between building something that works in a demo and building something people can trust.

What's the first thing that usually breaks in your AI-built apps?

#VibeCoding #AICoding

[Ganesh H N @ganesh2412]

That's a great approach, Abraham! A polished UI means little if the actual user journey breaks. I especially like testing what happens when a request fails. From my QA experience, recovery and data integrity often reveal more than the happy path. Do you also test what happens when users retry the same action?

[Ganesh H N @ganesh2412]

@ai_change_world Fair enough, Hugo! 😄 It's a solid experiment worth trying. I would be interested in comparing blind review time, defects caught and rework across different agents. Could make for a useful benchmark. Happy to exchange findings if either of us gives it a try!

[Ganesh H N @ganesh2412]

T3 Code, Cursor, Claude Code, Codex…

For your next comparison, give each tool the same bug, context and time limit.

Measure correctness, unrelated changes and review effort.

Which tool earned your trust—and what task earned it?

#AICoding #DevTools