Vikash Yadav

AI SuperBuilder @ Trilogy Innovations | IIT Jodhpur'24

[Vikash Yadav @vcossss]

@niveditjain @santanu_ai Warp is available for Ubuntu @niveditjain

[Vikash Yadav @vcossss]

@niveditjain Hi Nivedit,
I can feel the pain...

You should checkout https://t.co/WIenO1N5kx

And something similar: https://t.co/HSyvS1TwT1
(https://t.co/jjllG1YWsz)

[near @nearcyan]

this is how i claude code now. it's fun!

[Vinayak Sachdeva @vinayak_sachdev]

Love seeing innovation in this space! 🙌

We're tackling a similar problem with LearnLens (https://t.co/z2F3cOEvAg) — a Chrome extension that takes a different approach:

✅ Lives directly in YouTube (side-by-side panel, zero friction)
✅ Knowledge checks that test retention as you watch
✅ AI tutor for instant Q&A
✅ Math rendering & flow diagrams for technical content

Different philosophies on the same mission: making video learning actually work.

Always exciting when multiple teams push this space forward. Rising tide lifts all boats! Would love to connect and share learnings.

[Vikash Yadav @vcossss]

@nityaashah Yes, there will be personalized insights for students on the Timeback platform.

[Vikash Yadav @vcossss]

Not all hours behind a screen are learning hours💻
Some build knowledge, others slip into distraction.

TimeBack makes the difference crystal clear!
Fix your anti-patterns, and let every minute count⏳

Been cooking this for weeks—can’t wait to ship 🚀

[Vikash Yadav @vcossss]

@LamarDealMaker thanks Lamar!
It won't be standalone plugin, it will be the backend system powering @TimebackLearn

[Vikash Yadav @vcossss]

Know any Gemini video hacks I missed? 🙋
Drop them below 📝👇🏻

[Vikash Yadav @vcossss]

Using Gemini for video analysis?🎥
Want faster, cheaper, better results?✨

Here are 8 pro tips you probably didn't know 🧵👇🏻

[Vikash Yadav @vcossss]

Rule of Thumb 📌

1️⃣ Just one video, per prompt. Don’t pass multiple videos at once for best results. (Even though the limit is 10)

2️⃣ Place video before the prompt. Gemini works better when the video is the first element in the prompt, before any questions or instructions.

[Vikash Yadav @vcossss]

@UtsavBuilds @cursor_ai Planner mode is so useful! Will start using it right away..

[Vikash Yadav @vcossss]

1️⃣ Not sure about the correlation, haven't tested that. All videos I tried were of same compression ratio. Something to try.

2️⃣ Actually differences being filtered out surface the pesky dots. What I mean:

I have used findContours function to draw those red bounding boxes. This finds contours (edges) when there is sharp change in pixel values.

Now originally there were regions in the difference which had a gradual change of pixel values, so no contours identified there, filtering out the differences below a threshold caused these gradual hills to become steep cliffs, so now they are detected by the contours function resulting in those pesky dots.

These stiff cliffs when:
very small in area -> indicates leftover artifacts
over a certain area -> indicates actual change (activity on screen)

[Vikash Yadav @vcossss]

If nothing changes on the screen,
every frame of a screen recording should be identical... right?

Or is it? 🤨
Here’s how that assumption broke my detection system.

[Vikash Yadav @vcossss]

That's insightful @genSahil 💡

I remember facing a similar challenge in a project involved in making history videos using AI.

One potential issue we still might face is:
when we don't tell the model to pause, it sometimes starts the next sentence right after previous has ended. Cutting it manually there can make it sound abrupt.

Maybe we should:
⏸️ use <break time=...> where we need them and then
📏 extend the silences to exact lengths surgically using the transcription timestamps.

What do you think?

[Vikash Yadav @vcossss]

This has many use cases:
- User activity monitoring
- Automated UI Testing
- Video content analysis
- Attention tracking
- Security Systems

Has compression ever tripped you up? Let’s talk 👇

[Vikash Yadav @vcossss]

The system is now precise and intelligent!

Key Learnings:
🔍Always visualize — numbers alone can lie
🫧Video compression in common formats creates invisible differences