Sai Kiran Dev @DevSkg

Computer Vision and ML Engineer kiran.dev Frankfurt Joined October 2020

Tweets

614
Followers

83
Following

389
Likes

16K

Sai Kiran Dev @DevSkg

4 hours ago

@OfficialLoganK Good chase on efficiency and cost this was much needed. 3.1 pro felt more superior than 3.5 flash because of behaviour. I hope the behaviour is tuned like 3.1 pro.

0 0 0 79 0

View Details

Sai Kiran Dev @DevSkg

13 hours ago

Switching from Runpod to @modal is just a breath of fresh air. I can finally leave my OpenCode leave over night coding and testing.

0 0 1 19 1

View Details

We had a team of agents rebuild SQLite from its 835-page manual. It created a replica in Rust which passed 100% of a held-out test suite. Interestingly, cost varied 15x depending on which model mix we used.

583 846 12K 2.9M 4K

0 0 0 7 0

View Details

Sai Kiran Dev @DevSkg

14 hours ago

@gcduncombe @DecartAI Thats really cool, Do you any technique that can output the mesh data ?

0 0 0 90 0

View Details

Sai Kiran Dev @DevSkg

2 days ago

@AI_evangelist42 I use Opencode and it good

0 0 0 50 0

View Details

Sai Kiran Dev @DevSkg

2 days ago

@ajayyy_k @_skris Hostinger has

0 0 0 51 0

View Details

Sai Kiran Dev @DevSkg

2 days ago

@chrisoffner3d Amazing

0 0 0 264 0

View Details

Sai Kiran Dev @DevSkg

2 days ago

@andrew_n_carr How can we optimise the shape of the face to the model ?

1 0 1 107 0

View Details

Sai Kiran Dev @DevSkg

3 days ago

@max__drake WTF thats not India, this is India

1 0 3 84 0

View Details

Sai Kiran Dev @DevSkg

4 days ago

@PR0GRAMMERHUM0R @ajayyy_k 😂 seems relatable to us

1 0 0 245 0

View Details

Sai Kiran Dev @DevSkg

4 days ago

@ajayyy_k Hype around Iphone foldable and I guess

0 0 0 7 0

View Details

Sai Kiran Dev @DevSkg

4 days ago

Another perspective, Would you bet your $50,000 to win $1M with 50-50 chance ?

Saul @SaulWgmi

5 days ago

Which one do you prefer?

155 9 253 2.8M 23

0 0 1 28 0

View Details

Sai Kiran Dev @DevSkg

4 days ago

@beyarkay 😂

0 0 1 211 0

View Details

Sai Kiran Dev @DevSkg

5 days ago

@dragosroua @shadcn True 😂

0 0 2 1K 0

View Details

Kwang Moo Yi @kwangmoo_yi

6 days ago

Deng, Li, Qiu, et al., "Glob3R: Global Structure-from-Motion with 3D Foundation Models" Another one showing that you SHOULD do global optimization and bundle adjustment (BA) with Feed-forward geometry estimators. Traditional formula: keyframes + view graphs + tracks + BA.

4 13 97 6K 73

View Details

Sai Kiran Dev @DevSkg

6 days ago

@kwangmoo_yi Wow capturing slightly moving people would be better and handling reflection seems to be better. Exactly what I am looking for scanning cars. Waiting for the code release. Great work.

0 0 0 81 0

View Details

Yasser Dahou @dahou_yasser

4 months ago

We are releasing Falcon Perception, an open-vocabulary referring expression segmentation model. Along with it, a 0.3B OCR model that is on par with 3-10x larger competitors. Current systems solve this with complex pipelines (separate encoders, late fusion, matching algorithms). We developed a novel simpler "bitter" approach: one early-fusion Transformer (image + text from first layer) with a shared parameter space, and let scale + training signal do the work. Please check our work ! 📄 Paper: arxiv.org/pdf/2603.27365 💻 Code: github.com/tiiuae/falcon-… 🎮 Playground: vision.falcon.aidrc.tii.ae 🤗 Blogpost: huggingface.co/blog/tiiuae/fa…

26 166 989 121K 1K

View Details

Sai Kiran Dev @DevSkg

6 days ago

Man GLM5.2 is amazing for the cost.

0 0 0 12 0

View Details

Sai Kiran Dev @DevSkg

6 days ago

@soyposmoderno @Google Ideas > AI model to adjust this 3D mesh based on a person image > AI model to match the expressions from video to GNM > Take inspiration CAP4D and GaussianAvatar to make on device ultra realistic talking head avatar.