Utkarsh

731 posts

Utkarsh banner
Utkarsh

Utkarsh

@utk7arsh

incoming ml @nuro | @penn 27’ , @ucla 25’ | prev. @amazonscience | Researching on VLMs, VLAs | #ynwa

Philly Katılım Mayıs 2024
352 Takip Edilen1.2K Takipçiler
Sabitlenmiş Tweet
Utkarsh
Utkarsh@utk7arsh·
🏆 First Overall Winner at @Columbia University’s Devfest 2026 Feeling surreal to have my first grand win at a hackathon. We built Seamless, a service that turns streaming experience into shopping experience without disrupting the watching experience. Imagine buying the pizza you say in the show in real time without pausing or annoying ad breaks for it. Built using VideoDB for scene understanding and product placement, VideoPainter (open source) for product overlay, Snowflake for visual metadata storage, Dedaulus MCP for tool calling outlets like Kroger and DoorDash to recommend products based on user likings and location. Checkout the: Demo video: youtu.be/_pUlsVp3wR4 Devpost: devpost.com/software/seaml… GitHub: github.com/utk7arsh/Seaml… Now what do I do with this ps5 which is of no use to me lol. Please upvote on Devpost and star on GitHub if you like it 😄
YouTube video
YouTube
Utkarsh tweet mediaUtkarsh tweet mediaUtkarsh tweet media
Utkarsh@utk7arsh

WON THE HACKATHON!!!!!! Product post coming v soonnnnnnn!!!!! AHHHHHHHHHH

English
9
0
69
6.6K
Utkarsh
Utkarsh@utk7arsh·
@salomondrin Nuro is creating the tech for uber lucid partnership
English
0
0
0
53
Salomondrin 🤖
Salomondrin 🤖@salomondrin·
Excelentes noticias para Rivian, pero no se confundan... Lo más importante de estas asociaciones (incluyendo la de Lucid) es QUIÉN DE ELLOS PUEDE CREAR EL SOFTWARE DE PILOTO AUTOMÁTICO Y todos están kilómetros atrás de Tesla todavía e igual de pinches.
Rivian@Rivian

A fleet of R2 Robotaxis is coming exclusively to @Uber. ⚡🌿 Today, we announced a partnership to help both companies accelerate their autonomous vehicle plans across 25 cities in the US, Canada and Europe by the end of 2031. rivn.co/uber

Español
10
5
162
11.7K
Utkarsh
Utkarsh@utk7arsh·
I don’t understand this. Why is Uber partnering with R2 Rivian for 2031 when they already have L4 with Nuro and Lucid starting this year???
Rivian@Rivian

A fleet of R2 Robotaxis is coming exclusively to @Uber. ⚡🌿 Today, we announced a partnership to help both companies accelerate their autonomous vehicle plans across 25 cities in the US, Canada and Europe by the end of 2031. rivn.co/uber

English
3
0
7
170
Utkarsh
Utkarsh@utk7arsh·
@Av1dlive Glad I haven’t stopped using cursor
English
1
0
2
67
Utkarsh
Utkarsh@utk7arsh·
@Pseudo_Sid26 It’s a clickbait I’m sure (maybe I’m so wrong)
English
0
0
0
37
Avid
Avid@Av1dlive·
So I got early access to Minimax-M2.7 and it is insane this model is a builder's dream I made a space invader game on a raspberry pi here you go a look into it.
English
13
6
42
7K
Scorpihoee
Scorpihoee@cigs4breakffast·
Hekkiiii😍😍😍😍
IS
2
0
10
439
Utkarsh
Utkarsh@utk7arsh·
Inference engineering might as well end up on NY best selling this year. The book is gold. The lessons covered from the book is very well summarized in this article. Highly recommended as inference has application in every tech vertical right now
Avid@Av1dlive

x.com/i/article/2034…

English
1
0
10
437
Siddharth
Siddharth@Pseudo_Sid26·
Mamba-3 just dropped. You might be ignoring it as just another linear model but this could be the next real phase of State Space Models. Let's break this down: -State Space Models (SSMs) >These models don’t rely on attention >They treat sequences like a continuous dynamical system >Information flows through a hidden “state” over time >Everything is controlled by learnable matrices So instead of comparing every token with every other token (like attention), SSMs do something more physical. Workflow: Input sequence -> Linear projection -> State Space Layer (A, B, C matrices) -> Convolution (fast via FFT) -> Output projection - Mamba architecture it introduced selective state updates, meaning the model decides- >what to remember >what to forget and that too per token. -What Mamba-3 changes Mamba3 fixes the 3 biggest problems SSMs had: >Better recurrence (smarter memory updates) >Complex valued dynamics >MIMO (multi-input multi-output) Its like now we can actually use the mamba architecture efficiently. Read this paper to understand more, will drop a detailed article this weekend!!
Siddharth tweet media
English
9
2
40
1.3K
Utkarsh
Utkarsh@utk7arsh·
fun fact I learnt today: Niantic is using billions of 3D scans and images captured by Pokémon GO players to build a "living map" of the world, which is now being used to train navigation systems for autonomous delivery robots and drones. Pretty cool
English
0
0
11
179
Utkarsh retweetledi
Aarno
Aarno@TheGlobalMinima·
Been saying this for a year. Agentic AI is backend engineering far more than it is AI. This stands true for any technology, once you scale and abstract it enough, you’re only left with engineering problems. Learn > Event driven systems > Data pipelines > Distributed systems > API Design > Observability / monitoring
Ashutosh Maheshwari@asmah2107

x.com/i/article/2032…

English
45
110
1.1K
210K
Aryan Mangla
Aryan Mangla@Aryan_Mangla_·
Personal update: I won the Best Paper award in Robotics, Vision and Autonomous Vehicle - for my work in Robotics at NTech 2025 (NVIDIA’s biggest internal research conference) I became the youngest and the only sole author to ever receive this award at NVIDIA globally The whole journey to reach here was amazing and exciting at each step, really grateful for everyone’s support along the way!!
Aryan Mangla tweet media
English
43
27
400
17K
Utkarsh
Utkarsh@utk7arsh·
@Av1dlive Billing again goes brrrr. But tbh these attempts are just increasing the ease of switching between IDEs generally
English
1
0
2
38