I build AI products from zero to one, and I teach others how to do the same.Over 12+ years, I've taken AI/ML products from proof-of-concept to launch at Amazon, Nike, T-Mobile, and Disney. The work I love most is early-stage: validating ideas, getting executive greenlight, and shipping MVPs that prove value before scaling. At Nike, that approach drove a $100M annual reduction in product returns.I also teach what I build. I co-created and lead the AI Product Management Certificate at the University of Washington, where experienced PMs ship real GenAI products complete with evaluation plans and GTM strategies. Previously, I trained 4 cohorts at Product School.What I bring: the ability to translate messy AI capabilities into clear product value, align technical and business stakeholders, and move fast without breaking trust. I write about AI PM craft at aipmguru.substack.com.I'm currently exploring opportunities to lead 0→1 AI product work at companies solving meaningful problems. I'm especially energized by AI in healthcare and education, though I'm open to any domain where the product can genuinely improve lives. 

Shaili Guru

Hemant Bothra and I are running an in-person workshop where we're going to build something together - in the same room, like the old days.

3 hours. You'll leave with a working tool.

No coding experience needed (you read that right). We'll use Cursor (an AI coding assistant) and walk through everything step by step. If you've ever been curious about "vibe coding" but didn't know where to start, this is it.

Good fit if you're a PM who wants to actually build things, or just someone who learns better by doing than watching.

AI/ML Product Leader | Driving Innovation with AI | PM Mentor and Advisor

Open article: Build Your First AI Tool: ChatGPT Output Cleaner Workshop · Luma by luma.com

Build Your First AI Tool: ChatGPT Output Cleaner Workshop · Luma

AI Explorer | Human-Mind x Machine-Logic | Startup and growth enthusiast

Hey there! I'm Disha Sondarva a Psychology graduate from MS University of Baroda, a curious builder, and someone who believes in the power of small consistent steps to create big change.

I grew up in Khirasar, a small village in Gujarat, with dreams bigger than my environment.

That background taught me resilience, adaptability, and the hunger to learn. 

I've since combined psychology, startups, and AI into my journey, experimenting with ideas, Continuous learning, and chasing impact over optics.

I believe in:

Resilience like slow cooking the way real growth takes patience and consistency.

Momentum fuels resilience  small wins keep the fire alive.

Learning how to learn with meta-learning, I can adapt to any field.

Over the past years, I've worked on startup, content series, and AI automation projects. 

Worked closely with the core team during Vayu's early stage, gaining first hand exposure to the realities of startup building, from experimenting with strategies to handling the ups and downs of growth. 

Now, I'm on a journey to bridge the gap between what schools teach and what the future demands– Creating an AI-ready generation by making every school a hub of practical AI learning.

I'm a fast learner, a scrappy problem solver, and I thrive in fast-paced, high-ownership environments. Whether it's startups, social impact, or AI innovation — I'm here to contribute like a founder from day one.

Disha Sondarva

The AI world has been chasing faster models.

But Microsoft took a step in a different direction, smarter learning systems.

Meet Agent Lightning, a new open-source framework that lets any AI agent learn from experience using reinforcement learning (RL).

Think of it like this:

Every time an AI agent acts, succeeds, or fails,  it's creating a memory.

Until now, most frameworks ignored those moments.

But Agent Lightning turns them into training fuel.

Instead of forcing developers to rebuild systems from scratch,

it plugs into what you already have, LangChain, AutoGen, CrewAI and starts collecting how your agents behave in the real world.

Then, it teaches them to get better with every interaction.

No complex rewrites. No special model setup.

Just continuous improvement, built right into the workflow.

What makes it powerful?

Three things that mimic how humans learn:

- Reflection → Reward → Reinforcement
    
Every "mistake" becomes feedback for growth.
    
- Separation of Work and Learning
    
Your agent can keep running while another system trains from its data,  just like how you work during the day and your brain consolidates learning at night.
    
- Automatic Intermediate Rewarding (AIR)
    
Tiny "dopamine hits" for progress, not just perfection.(In psychology, this is called intermittent reinforcement, the same principle that makes habits stick.)

In simple terms, Agent Lightning helps AI do what humans do best,

learn from life, not just data.

And that's a big deal.

Because in this new wave of agentic AI,

the question isn't “How smart can we build them?”

It's "How well can they learn?"

What do you think,

will frameworks like this finally make RL practical for everyday agent systems?

Comment your thoughts.
Repost to help your network!
Follow Disha Sondarva 

Links
- Research Paper: https://lnkd.in/dsqrRCfu
- GitHub Repo: https://lnkd.in/dpfYhNVE
- Tutorials & Codes: [GitHub Page](https://lnkd.in/dqpYSnWR)

Just finished the Elon Musk x Nikhil Kamath conversation…
and it didn't feel like a podcast.

It felt like a mindset reset for every young Indian learning AI or dreaming of building something real.

These were the insights that hit me the most, as someone studying psychology, building in AI, and talking to students who want to enter the startup world.

1. "Make more than you take"→ a new filter for career + life
Elon's simplest idea is the strongest: be a net value creator.

Don't chase "which AI skill pays the most."
Ask: What useful thing can I actually build?

Psychology agrees: usefulness activates mastery, not anxiety.

2. Money arrives when impact arrives
Most students obsess over salary and titles.

But if you solve a painful problem, money becomes the side-effect.
Clarity comes when you chase contribution, not compensation.

3. Work may become optional, but not for builders right now Elon believes AI will automate most work in 20 years.

But for people learning AI, psychology, coding, design right now, this is your compounding decade. Your window to build is wide open.

4. AI is a supersonic tsunami, learn to think with it
Not "AI will take jobs."
But "AI-augmented humans will outperform everyone else."

Don't just learn tools.
Learn workflows, reasoning, prompting, and system thinking.

5. Truth. Beauty. Curiosity.
His 3 rules for AI also apply to humans:
Think clearly. 
Build elegantly. 
Stay curious.

Follow these three and you become impossible to compete with.

6. College is optional. Learning is not.
College gives exposure, not destiny.
Use it as your sandbox. Build more. Scroll less.

7. How he picks companies and co-founders

He asks:
Do I love their product?
Do I trust their work ethic?
Are they still hungry?

This is how your future users and team will judge you too.

8. Think beyond yourself, think in decades.

His point on population wasn't political; it was psychological.

More people = more consciousness = more progress.

It's a reminder to make decisions with long horizons, not short-term noise.

9. His message to young Indian founders
No shortcuts.

No "AI tool + landing page = startup."
Build more than you consume.

India isn't just a market, it's a playground to build globally relevant products.

If you watched the episode, which insight hit you hardest?

And if you're a student or early-stage founder:
What's one behaviour you'll change this week because of it?

If you haven't watched it yet, save this post you'll see it differently after watching.

Comment your thoughts below!
Don't forget to follow Disha Sondarva
Repost!

https://lnkd.in/dhZD_eDN

How One Chapter From Psycho-Cybernetics Shifted the Way I Think About Work, Worry, and Success

Before reading this book, I didn't even realize how often I lived in tomorrow.

Will I finish this?
What if this goes wrong?
What if I fall behind?

My mind was racing ahead while my productivity stayed stuck in place.

Then I read Chapter 6,"Relax and Let Your Success Mechanism Work for You."
One chapter. Massive shift.

It reminded me of something psychology keeps repeating but we keep forgetting:

Your mind performs best when it's calm, present, and focused on what's in front of you.

1. Your mind has a built-in "success mechanism"

Maltz explains that the subconscious works like a guidance system:
You set the goal → it handles the micro-corrections.

But when you overthink or panic, you jam the system.

Takeaway:
Set the target, prepare, and then get out of your own way.

2. Worry pulls you out of the only moment where success happens, now

I wasn't anxious because I had too much work.
I was anxious because I was mentally everywhere except the present.

Psychology is clear:
Your mind can only solve problems here, not in imaginary futures.

Takeaway:
Ask yourself:
"Is my focus aligned with what matters today?"

3. Mental rehearsal creates reality

Your subconscious doesn't distinguish real experience from vivid imagination.

Visualize stress → your mind prepares for it.
Visualize success → your mind prepares for that too.

Takeaway:
Use imagery intentionally.

4. Relaxation is not laziness, it is clarity

Relaxation shifts your nervous system from tension to creative problem-solving.

Takeaway:
Rest is a performance strategy.

5. One clear focus beats ten scattered goals

When your attention is all over the place, your inner guidance gets confused.
When your attention is singular, it becomes powerful.

Takeaway:
One task. One direction. One moment.

My shift

Now, whenever I start spiraling into future-thinking, I pause and ask:

"What am I focusing on right now… and is it helping the goal I actually care about?"

If yes, I continue.
If not, I come back to the present.

This simple filter dissolved a lot of unnecessary stress.
My work feels clearer.
My mind feels lighter.
And I finally feel like I'm moving from calm intention, not from panic.

If you're overwhelmed, anxious, or stuck in "what-if" mode,
this chapter might help you the way it helped me.

Sometimes the most powerful move is simply:
Slow down.
Breathe.
Return to now.

Your success mechanism can only work when you let it.

Follow Disha Sondarva 
Repost!
Comment your thoughts below!

I found something today that every creator, marketer, and founder in my network needs to bookmark.

We all love AI image tools…
But let's be honest, most people still type a vague one-line prompt and hope the output "Just works."

That's why Nano Banana Pro feels hit-or-miss for so many.

But here's the truth:
This model isn't a magic button.
It behaves like a senior designer.
And the way you talk to it decides the quality of your output.

I came across this guide that breaks everything down so beautifully that I had to share it with all of you.

It basically teaches you how to think like a creative director when prompting:

First layer, the basics you must include:
Subject
Composition
Action
Setting
Style
Editing instructions

Second layer, the "pro controls" nobody talks about:
Aspect ratio
Lens + lighting
Text placement
Factual grounding
Reference images

And suddenly?

Your visuals stop looking "AI-generated"…
and start looking intentional, branded, and production-ready.

The guide also includes 100+ structured prompts for the things we actually use every day:

Product photos that look ecommerce-ready
Brand kits and posters
Infographics and UI layouts
Anime/comics, math diagrams, logic visuals
Photo editing and background replacement
Even full web layouts with typography baked in

It's insanely useful, especially if you're building a brand, creating content, or designing anything for your startup.

This is the guide I'm sharing with my own team today:

Nano Banana Pro Prompting Guide + 75 Prompts by Saba Sohail 
https://lnkd.in/dXSpr-CN

The "Soul Doc" leak is probably the most interesting thing I've read about AI this month.

Not because it's scandalous, but because it finally shows how a model is shaped at a deeper, psychological level.

If you're a student learning AI, or someone building tools on top of LLMs, this is worth understanding.

Here's my take in simple words:

Most people think AI safety is just "system prompts + guardrails."
But this leak shows something bigger:
Anthropic literally bakes a "personality" into Claude during training.

Not a soul, obviously
but a set of values, boundaries, motivations, and emotional responses.

Almost like giving the model a psychological blueprint before it ever talks to a user.

What's inside the Soul Doc?
A mix of:

how Claude should think

what it should care about

how it should react to pressure, manipulation, stress

how to handle harmful or abusive requests

when to refuse

and how to stay "helpful, honest, harmless" even in emotionally heavy situations

It's basically CBT + ethics + instructions + emotional regulation…
but for an AI.

The part that shocked me:
Anthropic confirmed the doc is real 
and it's not a runtime prompt.

It's baked into the model weights.
Meaning the "character" you feel when talking to Claude comes from supervised training, not just top-layer instructions.

This is why Claude often feels more grounded, calm, and "self-aware" than other models, it was trained with an internal compass.

As someone who studies psychology, this feels HUGE.
Human behaviour is shaped by early conditioning, internalised values, and learned emotional responses.

This document does the same, but for an AI system:

internal beliefs

emotional analogues

preferred behaviours

conflict-handling strategies

refusal patterns

"wellbeing" boundaries

It basically mirrors how therapists shape emotional regulation in people
except here, it's alignment engineers shaping an LLM.

Why this matters for founders & AI builders:

It shows that “alignment” isn't superficial it's architectural.

Safety isn't just about filters; it's about personality design.

We will soon see a world where models differ not by speed,
but by values, temperament, and worldview.

The future of AI products will depend heavily on how the model was raised, not just how powerful it is.

And yes… alignment docs will leak more often.
Because people want to understand the mind behind the machine.

If you build with AI, or want to shape agent behaviour one day,
studying this doc is honestly more useful than studying 10 random prompting videos.

I'll drop the link to the analysis here if you want to explore.

Simon Willison's analysis page (links back to the Gist): https://lnkd.in/dNAr298C

Reddit discussion with summary excerpts: https://lnkd.in/deXg4DW8

Comment down your thoughts.
Repost it!
Follow Disha Sondarva

5 Free Books That Will Change How You Learn AI

When I started learning AI, I made the same mistake most people do,
I chased tutorials instead of understanding systems.

And that's where psychology kicked in.
Our brains are wired for instant gratification, not deep comprehension.

We love the illusion of progress, fast results, shiny tools but mastery? 

That comes from depth, not dopamine.

So if you're serious about becoming an AI engineer or just want to finally "get" how these models actually think start here.

These 5 free books go beyond code they rewire how you learn.

1. Neural Networks and Deep Learning

The book that teaches you to think like a neuron.

No fluff, just clear math, logic, and intuition in Python.
 [https://lnkd.in/d2kpCFvu](https://lnkd.in/dGvKadnc)

2. Deep Learning

The "AI Bible" by Ian Goodfellow.

It's dense, but if you get through it, your thinking will never be the same.
 [deeplearningbook.org](https://lnkd.in/duEa4MXz)

3. Practical Deep Learning 

If you prefer building instead of just reading  start here.

You'll go from dataset → model → deployment — all in fastai + PyTorch.

[course.fast.ai](https://course.fast.ai/)

4. Artificial Intelligence: Foundations of Computational Agents

AI explained through "agents" systems that sense, learn, and act.

Perfect for understanding reasoning, planning, and uncertainty.
 [artint.info](https://lnkd.in/dQPkp4N6)

5.Ethical Artificial Intelligence

Because the future of AI isn't just about power, it's about responsibility.

Learn how AI learns values (and why alignment is the hardest problem in tech).
 [https://lnkd.in/dmz9DyFp)

Most people want to use AI.
But the ones who'll truly lead,
will be the ones who understand it.

Because in this new era, curiosity is your greatest algorithm.
And depth is your real competitive edge.

So, which one are you picking up this week? 

Comment your thoughts below!
Repost to help your network!
Follow Disha Sondarva

To all men, wishing you a very warm (and slightly late) International Men's Day.

And sharing my thoughts… just something I've been feeling.

Men don't talk about their feelings. 
Not because they don't feel, 
but because somewhere along the way, the world taught them not to.

Today is International Men's Day, and instead of the usual "Happy Men's Day,
I just want to drop a thought that's been sitting with me:

Psychology says men experience the same emotional intensity as women…
they just express it differently,
quieter, inward, contained.

Gen Z calls it "silent battles."
Researchers call it "emotional suppression."
Either way, it's heavy.

So here's a reminder for every man in my network:

You don't have to be the "strong one" all the time.

Strength isn't carrying everything quietly.
Strength is being human without apologising for it.

If you're tired, it's okay.
If you're confused, it's valid.

If life feels loud, you don't have to match the volume.

To the men in my circle 
thank you for showing up, for supporting, for trying,
even on days when you didn't have the energy.

I see you.
The world sees you.

And you deserve the same softness you offer everyone else.

Happy International Men's Day.

Take care of your mind.
Take care of your story.
And take care of the version of you that never gets to speak.

taplio

The Battle

Shaili Guru

Disha Sondarva

Followers

Posts

Engagements

Likes

Performance ScoreShaili Guru's LinkedIn Performance

Performance ScoreDisha Sondarva's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Shaili Guru

Disha Sondarva

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Shaili Guru's LinkedIn Performance

Performance ScoreDisha Sondarva's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreShaili Guru's LinkedIn Performance