AI Architect & Builder - Inference, LLMs, SLMs GSoC Org Admin - Jenkins | Docker Captain | CNCF Ambassador
New Delhi
Joined December 2018
- Tweets7.1K
- Following1.3K
- Followers8.4K
- Likes37.6K
Pinned Tweet
What an amazing day! Presented a poster at @PyTorch Conference: how we optimised the RLM paper with pre-fix caching & batched Sub calls with @vllm_project
And a full house for our Pytorch Conference talk on running LLMs using Executorch on Android!
Shivay Lamba retweeted
Our latest model advances bring stronger capabilities for building and migrating software in low-resource programming languages such as COBOL, Fortran, and Visual Basic.
These languages power software that banks, governments, and businesses depend on. Progress here means understanding decades of business logic and helping developers modernize those systems while preserving what works.
We’re building these capabilities around the challenges developers and enterprise customers bring us. If your team is working on these systems, tell us which languages and workflows we should tackle next.
openai.com/index/low-resourc…
Shivay Lamba retweeted
How to cut coding-agent costs by ~78.5% using Jev to route between open models
All 32 selected coding tasks passed hidden tests.
Plus: Beam vs Muse Spark across 20 coding tasks - comparing cost, speed, and results 👇
Shivay Lamba retweeted
So cool to see @MLHacks giving out some @arduino UnoQs as part of prizes for their AI hackathon at @AllThingsOpen 🙌 – at Garner, NC
Shivay Lamba retweeted
Glad to be back at @AllThingsOpen after 2022! I am going to present a talk on how to improve security for running agents with @Docker Sandboxes - whether it is a coding agent or a multi agent system
Shivay Lamba retweeted
Continuous sensing creates memory bottlenecks on edge devices. Qualcomm & @ramin_m_h, Co-founder and CEO of @LiquidAI are advancing on-device AI with Liquid Context on @Snapdragon — compressing multimodal inputs for fast, low-power execution.
So cool to see @MLHacks giving out some @arduino UnoQs as part of prizes for their AI hackathon at @AllThingsOpen 🙌 – at Garner, NC
Glad to be back at @AllThingsOpen after 2022! I am going to present a talk on how to improve security for running agents with @Docker Sandboxes - whether it is a coding agent or a multi agent system
How to cut coding-agent costs by ~78.5% using Jev to route between open models
All 32 selected coding tasks passed hidden tests.
Plus: Beam vs Muse Spark across 20 coding tasks - comparing cost, speed, and results 👇
Shivay Lamba retweeted
Don’t forget about #PyTorchCon next week! We’ll be at booth D6 and hosting these events:
🔷 Keynote: MAX and Mojo: The Best Way to Extend @PyTorch — 10/20 @ 9:35 AM PT bit.ly/4yn0R0J
🔷 Demo: Scaling AI Across Devices: ExecuTorch — 10/20 @ 1 PM PT
bit.ly/4yjrAeG
An AI coding agent becomes useful when it can investigate a problem, change the code, test the fix, and follow through to a working deployment.
That’s what I wanted to explore with the newly launched @reflection_ai Beam 🚀
Beam is built for coding, reasoning and agentic workloads. KubeTrace AI puts those capabilities to work on @kubernetesio incidents, with inference through @nebiustf.
The agent can:
→ Investigate logs, events and application code
→ Trace a failure to its cause
→ Propose code or Kubernetes manifest changes
→ Run regression tests in a sandbox
→ Build and deploy the repaired application
→ Check whether it recovered
The execution layer checks every proposed change against policy, and hidden checks decide whether the incident is resolved. You can inspect the evidence, patches, test output and failed attempts.
Eating consistently while traveling is hard. Especially when you’re managing fatty liver and trying to be more mindful of your food choices.
Airport meals, hotel buffets, unfamiliar menus, and another promise to “get back on track when I’m home.”
So I started building Tend, an AI nutrition coach for people who travel ✈️🥗
Tend is still being developed, but I built an open-source version of the meal-analysis experience called Mealsight so you can try the vision + voice workflow yourself.
The idea: show the assistant your meal, talk about it, and ask follow-up questions.
How it works:
→ Upload a food photo or describe what you ate
→ The vision model analyzes the image and identifies the meal
→ Review what the AI identified
→ A confidence check asks for clarification when the meal is uncertain
→ Continue the conversation with questions about your food
It runs on @NVIDIAAI Nemotron 3.5 Super VL through @nebiustf powering the visual understanding and meal-related responses. The part I find most useful is being able to move from a food photo into a conversation without manually describing everything on the plate.
@GradiumAI brings in voice 🎙️
→ Speech-to-text for describing meals and asking questions aloud
→ Text-to-speech for hearing the assistant’s responses
Demo 🎥
youtube.com/watch?v=XgKR2wT7…
For people managing fatty liver, my goal is to make everyday food awareness easier to maintain while traveling, alongside guidance from their doctor or dietitian.
The meal analysis is something you can review and question. A photo alone won’t reveal every ingredient or exact portion.
The full Tend app is still in progress. For now, you can explore the open-source Mealsight version and build on it.
Code 👇 github.com/shivaylamba/meals…
Shivay Lamba retweeted
Muse Spark Open is on Nebius Token Factory from Day 0
It is the most capable US open-weight model to date.
Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.
Try it: tokenfactory.nebius.com/play…
Shivay Lamba retweeted
Today we’re releasing Muse Spark Open, the most capable US open-weight model to date.
Built for coding and agentic work, it excels at long-horizon tasks, complex software engineering, and multi-step reasoning.
Key Model Specs:
→ 816B total / 42B active parameters
→ 1M token context window
→ Natively multimodal across text, image, video, audio
