I create content about deep learning | The DevRel and Deep Learning Dude

Hacking on LLMs, curating multimodal datasets, and writing a book about RAG.

Harpreet Sahota 🥑

i've literally been waiting for years to have an OPEN SOURCE model like qwen3-vl-embedding

i tested its multimodal retrieval capabilities on a corpus of 412 short video clips and the results literally blew my mind

here are the queries i tested:

1. a cartoon guy drinks merlot wine

i like this query because we see how it can retrieve based on semantics (a cartoon), text (the label merlot), and temporal action in the video (the cartoon guy drinks the wine mid-way through the video)

2. garfield runs out of the door

semantics are awesome here, the model knows i'm talking about the cartoon character...

3. a woman in a blue dress sleeping on a red bench

the skeptic in you might think that it's retrieving just based on the red color block in the 2/3 of the video...but notice the specific part of the query "a woman in a blue dress..." and this is only shown in 3s out of the full 10s

this is such a huge release 

and it's gonna open up SO much more for multimodal video retrieval this year

on my wishlist is natural language search of pcd datasets, who gonna ship that?

you can hack around with the model using the resources below

check out the docs: here: https://lnkd.in/gh7yfBNP

and the quickstart nbhere:  https://lnkd.in/gbinsx7p

🤖 Hacker-in-Residence @ Voxel51| 👨🏽‍💻 AI/ML Engineer | 👷🏽‍♀️ Technical Developer Advocate | Learn. Do. Write. Teach. Repeat.

for those who missed it, or were there in-person and know the trouble i was facing 😆, my ODSC webinar is now live on youtube!

in this session i discuss visual ai techniques for automotive damage detection. 

i used the cardd dataset, which is a public dataset built specifically for vehicle damage analysis and demonstrated how fiftyone helps you explore, visualize, and build models to detect dents, scratches, cracks, glass shatter, tire flats, and broken lamps.

i feel like its a super practical workshop with some insightful workflows for visual AI in automotive settings, but as the creator i might be biased.

only one way to find out for yourself, watch the webinar!

https://lnkd.in/ewfz5vZD

Open video: Mastering Visual AI with Vision-Language Models & Advanced Evaluation Techniques by Harpreet Sahota on youtube.com

Mastering Visual AI with Vision-Language Models & Advanced Evaluation Techniques by Harpreet Sahota

sometimes i look back at the shit i get done and i genuinely impress myself

how teams really build Visual AI in the wild? what's the he messy reality?  

i don't know...and maybe its because nobody’s asking practitioners and compiling their answers so we can find out the landscape actually looks like. 

so, we at Voxel51 are kickin off the...

wait for it...

 State of Visual AI survey...vey...vey

for those out ther building visual AI systems i want your hot takes

what’s the hardest part about making visual ai a reality? 

let us know in this survey

it'll take you 30 minutes and is a much better use of time than baby sitting your runs and staring at your loss curves

i can't offer much beyond my heartiest thanks, but you will get early access to the full report with benchmarks, common pain points, and where you stand within the landscape. 

here's a link: https://lnkd.in/gnsADACy

MedGemma 1.5 supports detection, but for best results, you'll need to fine-tune. 

there's a kaggle competition using the model, i created a starter notebook to give you a jump start on how to fine-tune it for detection

checkout the starter notebook and additonal resources 👇🏼

cc Omar Sanseviero Megan Risdal Logan Kilpatrick

looking forward to the next meetup in boston!

Join our in-person meetup at the Microsoft Research Lab at the Massachusetts Institute of Technology in Boston, Massachusetts on Feb 18 to hear talks from experts on cutting-edge topics across AI, ML, and computer vision. Register here: https://lnkd.in/gi4kpSMB

Talks will include:

• SNAP: Towards Segmenting Anything in Any Point Cloud - Hanhui Wang at Northeastern University

• Culturally Adaptive AI - Anku R. at Massachusetts Institute of Technology

• Data Foundations for Vision-Language-Action Models - Harpreet Sahota 🥑 at Voxel51

• Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models - Sharut G. at Massachusetts Institute of Technology

• Neural Radiance Fields for Image Verification - Sage Simhon at Pasteur Labs & ISI

#computervision #ai #artificialintelligence #machinevision #machinelearning #datascience

Syed Abdul Sami Rizvi, Matthew Ebisu and 118 other attendees

View event: Feb 18 - Boston AI, ML and Computer Vision Meetup

Feb 18 - Boston AI, ML and Computer Vision Meetup

in the asu area and wanna get started with open source contributions? then don't miss this hack

On March 21, 2026, we’re hosting Agents World: Visual AI Hackathon @Arizona State University, together with Pavan Turaga 🚀 

This is a hands-on day: meet people, build with AI, eat pizza, and leave with something you created after a short workshop and a full day of hacking. 
You’ll work with the FiftyOne ecosystem and real models used in practice, like Qwen, Gemini Vision, Sharp, and more. 

If you’re an ASU student who wants to build, learn by doing, and work with real Visual AI tools, this is for you.

Registration is open:
https://lnkd.in/gBCHW5Fg

Hope to see you at ASU. I’ll be there with the amazing Voxel51 team:  Sid Mehta, Ritchie Martori, Jimmy Guerrero ! 🙌

Inspired by the amazing Hugging Face’s 2025 AI Timeline, we created one for the Fiftyone community. The FiftyOne Community Development Timeline highlights more than 131 contributions from 2025, including datasets, models, plugins, and integrations.

You can filter by type or modality and jump straight to each integration.
https://lnkd.in/ev2kpCGN

Big thanks to the entire Voxel51 community for building all of this together.
Harpreet Sahota 🥑, Daniel Gural, Paula Ramos, PhD, Jimmy Guerrero, Derek Van de Streek, Nicholas Lotz 🚀

Gonna be a great series of talks, lots of technical depth!

Chief Community Officer @ Voxel51 | Startup Advisor

Join us on Feb 18 for the Boston AI, ML and Computer Vision Meetup! Pre-registration is mandatory to clear building security - https://lnkd.in/gHtBuH49

Talks will include:

* SNAP: Towards Segmenting Anything in Any Point Cloud - Hanhui Wang at Northeastern University

* Culturally Adaptive AI - Anku R. at MIT

* Data Foundations for Vision-Language-Action Models - Harpreet Sahota 🥑 at Voxel51

* Better Together: Leveraging Unpaired Multimodal Data for Stronger Unimodal Models - Sharut G. at MIT

* Neural Radiance Fields for Image Verification - Sage Simhon at Pasteur Labs & ISI

***********

Want to build better computer vision models? FiftyOne is an open source toolkit from Voxel51 (our Meetup sponsor) that helps you curate datasets, evaluate model performance, visualize embeddings, catch annotation errors, and eliminate duplicate images—all in one place.

“pip install fiftyone” is all it takes to get started - https://docs.voxel51.com/

#computervision #ai #artificialintelligence #machinevision
#machinelearning #datascience #physicalai

I enjoy creating the future. You can learn more about me and my work at https://sadiestlawrence.com

Speaking Bio:

Sadie St. Lawrence is the Founder and CEO of the Human Machine Collaboration Institute (HMCI). HMCI pioneers research and education, to optimize human and machine interaction in the new era of knowledge work with AI, serving leading public and private organizations. 

Prior to HMCI Sadie founded Women in Data™, a non-profit organization with representation in 55 countries, and a community of over 70,000 data professionals. Women in Data earned recognition as a Top 50 Leading Non-Profit and the foremost community for Women in AI and Tech in 2021.

With three degrees in piano performance, psychology, and data science, Sadie's journey from neuroscience research to over a decade in data science and AI strategy showcases her diverse expertise. Her accolades include being named one of DataIQ’s Top 100 Most Influential People in Data (2024), Dataleum’s 30 Outstanding Women in Data (2023), Top 30 Women in AI (2022), Top 10 Most Admired Businesswoman (2021), Top 21 Influencer in Data (2021), and Top 30 Most Inspiring Women in AI (2021). She has also received the Outstanding Service Award from UC Davis (2019).

Additionally, Sadie's leadership extends to her role on the White House Council for Equitable Data and AI Training and educating over 700,000 individuals through her data science and AI courses at UC Davis, Coursera, and LinkedIn Learning.

Sadie Lawrence

Want to start the year off with a new job? I’m hiring for two roles and would love to meet people who are excited about building the future of work with AI.

1️⃣ AI & Technology Strategy Consultant
You’ll work directly with clients on AI strategy, assessments, and roadmaps, translating emerging tech into real business impact.
View here: https://lnkd.in/g8dj26BB

2️⃣ Social Media & Brand Manager
You’ll help shape and scale our voice across HMCI and my personal brand, telling clear, thoughtful stories about AI, work, and impact.
View here: https://lnkd.in/gVuH23CM

Both roles are hands on, high trust, and ideal for someone who loves ownership, learning fast, and working on meaningful projects.

If this sounds like you, or someone you know, I’d love for you to apply or share 💛

Sadie St Lawrence

For 10 years, Women In Data™️ has been building, supporting, and celebrating women in data, AI, and technology.

On March 6, 2026, we’ll come together in Sacramento for the Women in Data Gala to:
💜 Celebrate a decade of impact
💜 Honor leaders through our Women in Data Awards
💜 Recognize the women and allies who have helped shape this global community

This evening is about more than a milestone.

It’s about the relationships, careers, and opportunities created over the last ten years, and the future we’re continuing to build together.

If Women in Data has ever been part of your journey, I’d love for you to join us.

👉 RSVP here: https://lnkd.in/gDJaE-W2

I can’t wait to celebrate with you.

Feeling incredibly grateful today.

Because of the generosity and thoughtfulness of this community, Becoming an AI Orchestrator made it into the Top 11 AI books on Amazon.

I want to take a moment to sincerely thank the people who took the time to read the book and leave reviews. Your words, reflections, and support truly mean more than you know ❤️

Please give each of these amazing people a follow 👇🏼
Kristi Kunworee Baishya 
Angele Labady
Raghda Elshafie
Flavia Saldanha
Jessica Tappi
Sarah Elahi, CDMP
Sharon Matthews
Amanda Long ✨ Long
Natasha Lalwani
Angela Cao
Jenny Jimenez
Vikas Goyal
Jon Krohn 

I’m deeply appreciative of each of you for helping bring this message forward.

I spend my days doing three very different things:

1️⃣ Building AI strategies with cities and organizations

2️⃣ Teaching people how AI actually works

3️⃣ Explaining to smart leaders why “just using ChatGPT” isn’t a strategy
The biggest surprise?

The technical learning curve isn’t the hard part.

The hard part is unlearning how we’ve worked for the last 20 years.
AI doesn’t reward busyness.

It rewards clarity.

And clarity is uncomfortable.

Anyone else feeling this tension in their work?

Not everything generated by AI should be published.

And not everyone should outsource their thinking.

If AI removes your judgment, your perspective, or your accountability,
then it’s not augmentation.

It’s abdication.

The best AI users I know are sharper thinkers, not quieter ones.

No alternative text description for this image

#11 in AI books on Amazon - WOW THANK YOU 🥰 

My book has been out a little over a month and has stayed in the top 20 AI books on Amazon - so grateful for all the support of my work and reviews, thank you, thank you!

You can check out my book here: https://lnkd.in/gQuTz9ir

One thing I’ve noticed working with strong AI users:

They don’t ask AI to replace their thinking.

They ask it to:

🤓Stress-test ideas
🤓Surface blind spots
🤓Expand possibilities

Then they decide.

AI doesn’t make them quieter.

It makes them more intentional.

That difference matters more than most people realize.

Really honored to have Eric Weber join the Women In Data™️.

Eric has been a consistent supporter of Women in Data from the very early days, starting with the San Francisco chapter. He also personally inspired me early in my own data science journey as one of the first data scientists openly blogging, teaching, and sharing what was possible in this field.

We’re so excited to have his experience and perspective on the board and can’t wait to bring that expertise to our global community. Welcome, Eric 💜

Eric Weber is joining us as a Women in Data™ Board Member and we couldn't be more excited!


Eric is the Head of Data at Superhuman, where he leads data strategy and decision-making at the intersection of product, growth, and user experience.

Eric has been part of the Women in Data journey from the very beginning. As an early supporter of WiD, he sponsored the San Francisco Chapter, helping lay the foundation for the community we continue to grow today.

We’re incredibly grateful for Eric’s long-standing belief in our mission and excited to have his perspective, leadership, and continued support at the board level. Welcome, Eric!

AI ecosystems don’t happen by accident. They’re designed.

In the City of Rancho Cordova, we’re intentionally building one that connects sustainability, education, and research into a system that actually works in the real world.

Our focus is simple and disciplined.

Sustainability through our IR² model, so innovation can fund itself long term.

Education with NVIDIA’s Deep Learning Institute, so AI skills turn into real jobs, not just certificates.

And research and innovation with our university partners, so we keep advancing what’s possible at the intersection of humans and machines.

This isn’t about hype. It’s about infrastructure, talent, and execution.

Proud of what we’re building with the City of Rancho Cordova and our partners. And we’re just getting started.

https://lnkd.in/gG4CX3ie

N8N agent node is a game changer in your workflow. It allows you to choose the model, memory and tools you want your agent to use. 

Check out the full tutorial on my YouTube channel.

If you’re a leader trying to “roll out AI” and feeling stuck, start here 👇🏼

Ask your team three questions:
1️⃣ What decisions do you make repeatedly?
2️⃣ What work feels manual but high-stakes?
3️⃣ Where do errors actually cost us money or trust?

THEN introduce AI.

Not the other way around.

Tools don’t fix broken workflows.

Good orchestration does.

I love this announcement from Amazon Web Services (AWS), because it mirrors something we learned a long time ago in neuroscience.

In the human brain, efficiency comes from specialization. You don’t ask your visual cortex to manage memory storage, and you don’t force motor neurons to hold context they don’t need. When systems are overloaded with responsibilities they weren’t designed for, performance drops and failure rates go up.

That’s exactly what’s been happening in Spark for years.

We’ve been asking computers to also manage storage, capacity planning, and fault tolerance. The result? Overprovisioning, wasted spend, brittle jobs, and failures driven by data skew rather than logic.

With Amazon EMR Serverless storage, AWS essentially “rewired” this pathway.

Amazon EMR Serverless has essentially rewired this pathway by eliminating local storage provisioning for Apache Spark workloads.

By decoupling storage from compute, shuffle data is handled by a dedicated, elastic layer. 

Compute is used when needed and released immediately when done.The system adapts dynamically, just like a well-functioning brain.

The outcome is real-world impact, not just cleaner architecture:
🤓Up to 20% reduction in data processing costs
🤓Fewer Spark job failures from disk constraints and skew
🤓Faster scale-down between stages, especially in funnel-shaped queries
🤓Engineers spending time on insights, not infrastructure tuning

This is one of those changes that looks “under the hood” but fundamentally shifts how teams think about building analytics at scale. Less cognitive load for humans, better resource allocation for machines.

If you care about cost efficiency, resilience, and designing systems that actually behave intelligently under pressure, this is worth your time.

Read the full announcement here and see how serverless storage changes the Spark model:
https://lnkd.in/gHuiv_MT

#AWSPartner

taplio

The Battle

Harpreet Sahota 🥑

Sadie Lawrence

Followers

Posts

Engagements

Likes

Performance ScoreHarpreet Sahota 🥑's LinkedIn Performance

Performance ScoreSadie Lawrence's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Harpreet Sahota 🥑

Sadie Lawrence

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);}Harpreet Sahota 🥑's LinkedIn Performance

Performance ScoreSadie Lawrence's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreHarpreet Sahota 🥑's LinkedIn Performance