Feyn (@FeynAI) header
Feyn (@FeynAI) profile picture

Intelligence shaped by you

San Francisco
Joined December 2024
Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)Photo by Feyn (@FeynAI)
Feyn retweeted
Building FeyNoBg also exposed a tooling problem. Image matting models usually live in isolated repos with different interfaces. So we built NoBg, a Python library for running, training models, and evaluating image mating models. Currently, it supports BiRefNet with more architectures coming. We hope you use it to build something exciting. github.com/feyninc/nobg
1
1
4
237
Happy to share that @ChonkieAI just crossed 4K GitHub stars and 4 million PyPI downloads! 🎉 from a small side project to a fast growing library used by millions. Huge thanks to the community, so stay tuned as bigger things are coming 🔥
1
6
331
Our OSS engineer @itsclelia recently built 𝗹𝗶𝘁𝗲𝘀𝗲𝗮𝗿𝗰𝗵, a fully local document ingestion and retrieval CLI/TUI application powered by LiteParse ⚡ litesearch demonstrates how developers can assemble a high-performance, local-first retrieval pipeline using open tools from across the ecosystem: • Parsing: LiteParse, the fast and accurate document parser we recently open sourced • Chunking: @ChonkieAI • Embeddings: A local @nomic_ai model via @huggingface transformers.js • Vector storage: A local @qdrant_engine edge shard (custom-built in Rust and compiled as a native add-on) • Retrieval: Query stored files with optional path-based filtering and configurable relevance thresholds • Runtime: @bunjavascript for speed and versatility 💻 Check out the repository and try it yourself: github.com/AstraBert/litesea… 📚 LiteParse docs: developers.llamaindex.ai/lit…
15
2
78
14,938
Feyn retweeted
stop using MCPs to give your agents documentation. mandex gives your AI agent the same docs — but local, version-pinned, and 25ms instead of 800ms. One download. Works offline. No API keys. No rate limits. No monthly bill. 100+ packages. 1,500+ versions. Free forever. 🔗👇
5
7
1
36
3,207
Glad to announce we have crossed 2 million downloads on our library at @ChonkieAI
1
1
11
433
Deep research stories #2: People using deep research to understand manga
1
6
220
Feyn retweeted
i’ve been using this for the past 4 days and have been blown away at the quality information the agent gets for me 10/10 must try
Today we are launching Chonkie Labs. Deep research on autopilot. Here is a dashboard we generated on Anthropic's enterprise moves. Four Salesforce hires, a $200M Snowflake partnership, 30K person Accenture training pipeline, and the default model in ServiceNow. All of this points to one pattern: a $26B play for enterprise SaaS. Generated in 3 minutes. labs.chonkie.ai/share/7852db…
2
3
52
13,187
Introducing 𝗿𝗮𝗴-𝗿𝘀, a Rust-native implementation of the RAG stack🦀 rag-rs allows you to load data from PDF, plain-text and markdown documents to @qdrant_engine, using PdfExtract for parsing, memchunk by @ChonkieAI for lightning-fast chunking and BM25 for creating sparse embedding🗄️ Once you loaded your vector collection, you can serve it as an API endpoint, with out-of-the-box support for CORS, rate limiting and logging💻 Client request → Embedding → Retrieval → Response generation with any @OpenAI model✅ Take a look at the demo👇 Install the alpha release with 𝘤𝘢𝘳𝘨𝘰 𝘪𝘯𝘴𝘵𝘢𝘭𝘭 𝘳𝘢𝘨-𝘳𝘴@0.2.0-𝘢𝘭𝘱𝘩𝘢 and find the repo on GitHub 👉 github.com/AstraBert/rag-rs
1
1
7
708
day 3: work + life logs > revised and solved LeetCode questions on binary tree traversals (inorder, preorder, postorder) > completed reading the @ChonkieAI documentation tried out the code examples that were in it, and it’s surprisingly very easy to implement
1
1
4
181
How fast is too fast for chunking?
2
1
12
745
i like
The embedding API ecosystem is a mess. Every SDK has different types. No retry logic. No cost tracking. Pricing scattered across a dozen docs pages. We built 🐱catsu to fix this. A unified client for 50+ embedding models across 11 providers.
2
1
6
595
Chonkie and Weaviate 1. Weaviate Handshake: Use Chonkie’s `MarkdownChef` to extract chunks and ingest them into Weaviate to search (github.com/weaviate/recipes/…)
1
4
1
13
1,583
Feyn retweeted
Stop the API ping-pong. 🏓 With the new Pipeline API, you can build complex multi-chunking and refinement flows in the UI and execute them with a single call. Everything is initialized server-side, meaning zero back-and-forth and significantly lower latency. ⚡️ One request. That's all.
1
1
1
4
1,417
I'm glad to announce that we reached over 1,000,000 downloads at @ChonkieAI🔥 our library is growing, we reached over 3,000 stars on github, more than 250 repos are dependent on Chonkie and have been used by people from big tech companies like openai, microsoft and several others
2
1
22
1,026
I recently had to pick a RAG stack for an MVP, so I wanted something super hands off. I want control on the chunking strategy and then everything else might as well be a black box. I found @ChonkieAI , but was a little skeptical. Booked a call with the founder, @shreyash_nm, within 3 hrs, he spent an hour on the call with me, and answered questions even past midnight. I don’t know how well chonkie will do, but it did raise my bar for CX tremendously.
2
1
7
511
a millie a millie a millie a millie a miilie
6
1
26
1,106
Replying to @mattpocockuk
@ChonkieAI recently added full TS support to their SDK (they went Python first too).
2
3
570
Chonkie v1.4 is here. An open-source AI ingestion framework built for smarter document parsing and RAG pipelines. This update introduces Pipelines for retrieval workflows, Chefs for cleaning messy data, TableChunkers for structured markdown, and Handshakes to connect with databases like Chroma, Qdrant, Weaviate, and PostgreSQL.
5
62
444
30,445
Summer internship wrap up, signed: AI Agents and Graph RAG! @ChonkieAI @MistralAI @googlecloud @neo4j I spent three months bundling these technologies in a python backend to help give a push to phrama tech! More on this: linkedin.com/posts/mohamedal…
2
1
6
275
Amazing things are happening at @ChonkieAI
5
2
21
1,317