nerdai @_nerdai_

Applied ML Scientist @vectorinst ◦ Ex Founding Engineer @llama_index ◦ PhD (@UWaterloo) ◦ CIPT github.com/nerdai Toronto Joined November 2020

Tweets

944
Followers

2K
Following

1K
Likes

3K

Logan Markewich @LoganMarkewich

2 months ago

When Massi joined a year ago, we met up in SF for a week for onboarding. Within that week, the idea for workflows was hatched, and since then, has become my favourite feature to both use and maintain. It's easier than ever to get started and contribute! Check it out! 🚀🦙

LlamaIndex 🦙 @llama_index

2 months ago

2 18 105 41K 70

Download Image

0 2 8 615 0

Min Choi @minchoi

2 months ago

RIP Cursor/Windsurf Claude Code for VSCode 🤯

402 756 11K 1.2M 7K

Download Image

Jerry Liu @jerryjliu0

2 months ago

Our team is actually cracked at document parsing. I threw in an old equity research report on Amazon, and watched as our parsing agent rendered an entire combination chart (3 graphs in 1) into a 2D table with clean accuracy. And it interleaved it perfectly with the other…

11 20 194 27K 238

Download Video

nerdai @_nerdai_

3 months ago

🎉 📣 Very excited to share that our submission to the first-ever CODEML Workshop at this year's @icmlconf 2025 has been accepted! About the workshop: CODEML, standing for Championing Open-source Development in Machine Learning, is a new workshop that aims to recognize…

1 3 24 8K 17

Download Image

nerdai @_nerdai_

3 months ago

📣 FedRAG now supports a bridge to LangChain—thanks to an external contributor! With FedRAG, you can assemble your RAG system and fine-tune your generator/retriever component models to better adapt to the contents of your knowledge store. After fine-tuning your system, you can…

0 0 2 14K 1

Download Image

nerdai @_nerdai_

3 months ago

🚀 Introducing new core abstractions in FedRAG—NoEncode RAG with MCP! With traditional RAG, there are three components: retriever, generator, and knowledge store; and, knowledge contained in the knowledge store has previously been encoded by the retriever model and stored for…

6 13 69 96K 16

Download Image

nerdai @_nerdai_

4 months ago

🎉🚨 New Integration Alert! I'm happy to share that FedRAG now supports Unsloth! You can now build RAG systems using any of @UnslothAI's FastModels as your generator, then fine-tune with all of Unsloth's performance accelerators and patches. I'm thrilled to finally integrate…

0 0 12 538 6

Download Image

nerdai @_nerdai_

4 months ago

⭐️ FedRAG Updates: New Version, Evals & Benchmarking! ⬢ Version 0.0.19 Released: Featuring streamlined import patterns and our new evals module ⬢ Benchmark Your RAG Systems: Now test performance against popular benchmarks like MMLU ⬢ New Integration Notebooks: Added tutorials…

0 0 6 265 1

nerdai @_nerdai_

4 months ago

🎉 New FedRAG Tutorial Notebook + Vid! Following up on our introduction announcement to FedRAG, I'm happy to share the first of several notebook tutorials + accompanying YouTube videos to get you started with the library. We've also added a dedicated "Notebooks" section to our…

4 7 44 78K 13

Download Image

nerdai @_nerdai_

4 months ago

🚀 Very Excited to Introduce FedRAG! Today, I'm pleased to share my latest open-source project that I have been working on since joining Vector Institute this past January. FedRAG is a framework for fine-tuning RAG systems. It endeavours to simplify the fine-tuning of RAG…

7 19 71 14K 71

Download Image

Wenqi Glantz @wenqi_glantz

4 months ago

Happy to share my first NVIDIA Tech Blog post! Inviting you to explore the open-source NVIDIA Agent Intelligence Toolkit, a brilliant resource for building agentic applications. Be sure to add it to your toolkit! bit.ly/438WQzs

3 9 45 13K 31

nerdai @_nerdai_

4 months ago

💥 🚀 Exciting News! After a few months of diligent work, I'm thrilled to announce the pilot launch of AI Pocket References—from the AI Engineering team Vector Institute! These pocket references are a collection of streamlined information spanning a wide array of areas in AI. In…

0 1 9 11K 5

Download Image

Jerry Liu @jerryjliu0

5 months ago

We’re excited to provide an official A2A document agent that can parse a complex, unstructured document (PDF, Powerpoint, Word), extract out insights from it, and pass it back to any client. The A2A protocol allows any compatible client to call out to this agent as a server. The…

8 69 460 34K 555

Download Image

nerdai @_nerdai_

6 months ago

I built my first MCP Server (@AnthropicAI) this past week, Goodness MCP. It delivers good news to MCP Clients like Claude Desktop! Sharing a new @Medium blog post providing a walkthrough of it as well as my thoughts on the overall experience. nerdai.medium.com/goodnews-mcp-g…

1 0 2 376 0

Download Image

nerdai @_nerdai_

6 months ago

Positive news at your fingertips via Goodnews MCP! Using @AnthropicAI's MCP, Goodnews exposes a `fetch_good_news_list` tool, which brings headlines from various categories to MCP clients. Check out the Github repo to hook it into your Claude Desktop! github.com/VectorInstitut…