[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]

@akshay_pachaar Avatar @akshay_pachaar Akshay 🚀

Akshay 🚀 posts on X about mcp, llm, inference, token the most. They currently have XXXXXXX followers and XX posts still getting attention that total XXXXXX engagements in the last XX hours.

Engagements: XXXXXX #

Engagements Line Chart

Mentions: XX #

Mentions Line Chart

Followers: XXXXXXX #

Followers Line Chart

CreatorRank: XXXXXXX #

CreatorRank Line Chart

Social Influence #

Social category influence technology brands XXXX% stocks XXXX%

Social topic influence mcp #23, llm #30, inference 5%, token #611, github #98, protocol 2.5%, deepseekr1 2.5%, ollama 2.5%, scout 1.25%, meta XXXX%

Top accounts mentioned or mentioned by @avichawla @jorgelison @cheqdnetwork @opendemind @akshaypachaar @ggerganov @motiadev @firecrawldev @ollama @crewaiinc @zepai @cometml @milvusio @mistralai @robinchps @jendor_skii @marshalkal @pbo_llp @aniksingal @autoppiaai

Top assets mentioned Microsoft Corp. (MSFT)

Top Social Posts #

Top posts by engagements in the last XX hours

"Meta used co-distillation to train Llama X Scout and Maverick from Llama X Behemoth. Of course during the initial stages soft labels of the Teacher LLM won't be accurate. That is why Student LLM is trained using both soft labels + ground-truth hard labels"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:24 UTC 215.4K followers, XXX engagements

"Model Context Protocol (MCP) clearly explained:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-06-13 12:34:10 UTC 215.4K followers, 1.1M engagements

"5 Define Response Synthesis Agent Synthesis Agent serves as final quality control layer refining retrieval outputs from specialized agents into polished user-friendly responses. Check this out 👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-23 13:20:42 UTC 215.4K followers, 1054 engagements

"Transformer vs. Mixture of Experts in LLMs clearly explained (with visuals):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:03 UTC 215.4K followers, 234.4K engagements

"4 LlamaCPP (the OG) @ggerganov 's LlamaCPP enables LLM inference with minimal setup and state-of-the-art performance. Here's DeepSeek R-1 running on a Mac Studio 🪄"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-02-07 13:06:03 UTC 215.3K followers, 8728 engagements

"MCP allows AI agents to connect with external tools and data sources through a plugin-like architecture. It's rapidly taking over the AI agent landscape with millions of requests processed daily. But there's a serious problem. 👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-20 13:34:50 UTC 215.4K followers, 15K engagements

"3 Co-distillation - Start with an untrained Teacher and Student LLM. - Generate softmax probs over the current batch from both models. - Train the Teacher LLM on the hard labels. - Train the Student LLM to match softmax probs of the Teacher. Check this visual 👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:19 UTC 215.4K followers, XXX engagements

"Challenge 1) Notice this pattern at the start of training: - The model selects "Expert 2" - The expert gets a bit better - It may get selected again - The expert learns more - It gets selected again - It learns more - And so on Many experts go under-trained"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:39 UTC 215.4K followers, 3475 engagements

"Let's build a multi-agent content creation system (100% local):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-16 13:02:18 UTC 215.4K followers, 116.5K engagements

"6 Setting up a query engine The query engine takes query string to use it to fetch relevant context and combines them using the prompt template before sending it to the LLM that generates final response The LLM used here is the latest Qwen3-Coder"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-24 17:07:24 UTC 215.4K followers, XXX engagements

"System overview: - User submits a query - Router agent identifies modality and triggers a specialist - Specialist agent sends relevant context to response generator - User receives a coherent response Let's dive into the code"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-23 13:20:35 UTC 215.4K followers, 2271 engagements

"KV caching is a technique used to speed up LLM inference. Before understanding the internal details look at the inference speed difference in the video: - with KV caching X seconds - without KV caching XX seconds (5x slower) Let's dive in"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-05-29 13:03:26 UTC 215.4K followers, 18.7K engagements

"Before we dive in here's a quick demo of what we're building Tech stack: - @motiadev as the unified backend framework - @firecrawl_dev to scrape web content - @ollama to locally serve Deepseek-R1 LLM The only AI framework you'll ever need to learn 🚀"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-16 13:02:19 UTC 215.4K followers, 3747 engagements

"@crewAIInc @zep_ai @Cometml Here's an overview of the system we're building - User sends a query - Assistant runs web search via MCP - Query + results go to Memory Manager - Memory Manager stores context in Graphiti - Response agent crafts the final answer"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-12 12:33:41 UTC 215.4K followers, 5204 engagements

"5 Creating a prompt template A custom prompt template is use to refine the response from LLM & include the context as well:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-24 17:07:20 UTC 215.4K followers, XXX engagements

"Let's build an MCP-powered financial analyst (100% local):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-05-27 14:31:23 UTC 215.4K followers, 563.8K engagements

"10 MCP AI Agents and RAG projects for AI Engineers (with code):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-18 12:52:07 UTC 215.4K followers, 267.6K engagements

"Drag-and-drop UI to build AI agent workflows Sim Studio is a lightweight user-friendly platform that makes creating AI agent workflows accessible to everyone. Supports all major LLMs MCP servers vectorDBs etc. XXX% open-source"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-06 12:31:37 UTC 215.4K followers, 92.3K engagements

"Universal Tool Calling Protocol A safer reliable and scalable alternative to MCP. Universal Tool Calling Protocol (UTCP) lets any agent talk to any toolHTTP gRPC WebSocket even your local CLIwithout wrappers latency or lockin. XXX% open-source"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-22 12:34:36 UTC 215.4K followers, 67.8K engagements

"10 GitHub repos to build a career in AI engineering. It covers: - Basics of AI/ML - Neural networks - Research paper implementations - MLOps - LLMs/RAG/Agents etc. Check the XXX% free roadmap in the thread below:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-18 07:36:32 UTC 215.4K followers, 67K engagements

"A Python decorator is all you need to deploy AI apps Beam is an open-source alternative to Modal for deploying serverless AI workloads. Simply pip install beam add a decorator and turn any function into a serverless endpoint XXX% open-source"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-08 12:42:00 UTC 215.3K followers, 39.4K engagements

"Temperature in LLMs clearly explained (with code):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-07 12:36:35 UTC 215.3K followers, 264.1K engagements

"10 GitHub repos that will set you up for a career in AI engineering (100% free):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-17 12:32:17 UTC 215.4K followers, 251.8K engagements

"First of all what is Pixeltable Pixeltable is a go-to Python library for Multimodal AIstreamlining entire pipeline from data storage to model execution. Handles images videos text & audio effortlessly. Our MCP servers are built on top of Pixeltable"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-23 13:20:34 UTC 215.4K followers, 3098 engagements

"I just built the ultimate MCP server for Multimodal AI. It lets you do RAG over audio video images and text XXX% open-source here's the full breakdown.👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-23 13:20:32 UTC 215.4K followers, 105.9K engagements

"MCP is on fire. AI agents can now talk to real world tools apps and actually get stuff done. This changes everything. Here are XX amazing examples:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-11 12:28:46 UTC 215.4K followers, 272.1K engagements

"MCP is on fire. AI agents can now talk to real world tools apps and actually get stuff done. This changes everything. Here are XX amazing examples:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-04-11 12:30:08 UTC 215.4K followers, 717.8K engagements

"Let's build an MCP server that connects to 200+ data sources (100% local):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-05-30 14:05:56 UTC 215.4K followers, 413.4K engagements

"AI Engineering Hub just crossed 10k GitHub stars Its XXX% open-source and hosts 70+ free hands-on demos. Here are XX MCP RAG and AI Agents projects for AI engineers:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-06-20 12:30:36 UTC 215.4K followers, 287.8K engagements

"Microsoft just changed the game 🔥 They've open-sourced bitnet.cpp: a blazing-fast 1-bit LLM inference framework that runs directly on CPUs. Why is this a game-changer❓ You can now run 100B parameter models on local devices with up to 6x speed improvements and XX% less energy consumptionall without a GPU The future we've been waiting for: fast efficient and private AI that works anytime anywhere.✨ The model is also available on HuggingFace you can try it right away I've shared Link to the GitHub repo in next tweet where you'll find all the details. _____ Find me @akshay_pachaar ✔ For more"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-05-12 12:45:48 UTC 215.2K followers, 525.1K engagements

"2 Hard-label distillation - Use the Teacher LLM to get the output token. - Get the softmax probs. from the Student LLM. - Train the Student to match Teacher's output. DeepSeek-R1 was distilled into Qwen & Llama using this technique. Check this visual 👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:15 UTC 215.4K followers, 1435 engagements

"MLOps feels broken when it comes to AI agents ZenML bridges the gap between classical ML and modern AI giving you one unified framework for your entire AI stack. Here's why it's revolutionary: 🔄 Same MLOps principles for models AND agents ⚡ Pipeline everything from sklearn to LangGraph 🎯 Data-driven agent architecture decisions 📊 Full observability with LiteLLM & Langfuse 🗂 Version prompts like you version datasets But that's not all: 💬 Chat with your pipelines using natural language 🔧 Wrap existing code with simple @ step decorators ☁ Deploy locally or scale to Kubernetes 🤝 Integrates"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-19 16:30:00 UTC 215.4K followers, 16.8K engagements

"I have tested 100+ MCP servers in the last X months Here are X must-use MCP servers for all developers (open-source):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-06-28 13:36:33 UTC 215.4K followers, 538.8K engagements

"Challenge 2) Some experts may get exposed to more tokens than othersleading to under-trained experts. We prevent this by limiting the number of tokens an expert can process. If an expert reaches the limit the input token is passed to the next best expert instead"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:47 UTC 215.3K followers, 2537 engagements

"1 MCP meets Ollama In this video you'll learn how build a XXX% local MCP client that you can connect to any MCP server. XXX% open-source code step-by-step guide:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-18 12:52:09 UTC 215.4K followers, 8256 engagements

"Say your vocab has 100k tokens and data has X trillion tokens. Storing softmax probabilities over the entire vocab for each input token needs 500M GBs of memory under fp8 precision. This is where we jump to our second technique .👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:09 UTC 215.4K followers, 1420 engagements

"To understand KV caching we must know how LLMs output tokens. - Transformer produces hidden states for all tokens. - Hidden states are projected to vocab space. - Logits of the last token is used to generate the next token. - Repeat for subsequent tokens. Check this👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-05-29 13:03:31 UTC 215.4K followers, 16.6K engagements

"We solve this in two steps: - Add noise to the feed-forward output of the router so that other experts can get higher logits. - Set all but top K logits to -infinity. After softmax these scores become zero. This way other experts also get the opportunity to train"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:44 UTC 215.4K followers, 2975 engagements

"A Crash Course on Building AI Agents Here's what it covers: - What is an AI agent - Connecting agents to tools - Overview of MCP - Replacing tools with MCP servers - Setting up observability and tracing All with XXX% open-source tools"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-12 12:32:00 UTC 215.4K followers, 113.3K engagements

"Containerized versions of 450+ MCP servers MCP Containers bundles hundreds of Model Context Protocol servers into ready-to-run Docker imagesno setup headaches no version drift. XXX% open-source and secure"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-13 12:43:15 UTC 215.4K followers, 92.3K engagements

"@_avichawla Connect any LLM to any MCP server + XXX% local MCP clients Thanks for sharing Avi 🙌"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-22 06:35:22 UTC 215.3K followers, 2152 engagements

"10 MCP AI Agents and RAG projects for AI Engineers (with code):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-19 07:30:00 UTC 215.4K followers, 82.8K engagements

"ML researchers just built a new ensemble technique. It even outperforms XGBoost CatBoost and LightGBM. Here's a complete breakdown (explained visually):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-14 12:34:24 UTC 215.4K followers, 399.8K engagements

"4 Indexing & storing Embeddings created by embedding model are stored in a vector store that offers fast retrieval and similarity search by creating an index over our data. We'll use a self-hosted @Milvusio vector database:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-24 17:07:16 UTC 215.4K followers, 1119 engagements

"MCP security is completely broken Let's understand tool poisoning attacks and how to defend against them:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-20 13:34:49 UTC 215.4K followers, 462.7K engagements

"Those were the three techniques to train one LLM using another. We discussed: - Soft-label distillation - Hard-label distillation - Co-distillation Here's the visual again for your reference 👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:29 UTC 215.4K followers, 3132 engagements

"4 MCP powered Agentic RAG In this project you'll learn how to build your own MCP server It connects to Cursor & performs deep web searches as well as RAG over a specified directory. XXX% open-source code with a step-by-step guide:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-18 12:52:23 UTC 215.3K followers, 4684 engagements

"The only AI framework you'll ever need to learn In this video we create an agentic workflow for writing and publishing content using Motiaa unified system where APIs background jobs events and agents are just plug-and-play steps. XXX% open-source"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-17 07:35:16 UTC 215.4K followers, 20.5K engagements

"Finally A Text-to-SQL tool that actually works Vanna is an open-source RAG framework for complex Text-to-SQL generation. It manages dynamic data and allows custom RAG model training for greater accuracy. XXX% open-source"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-15 12:32:51 UTC 215.4K followers, 99.4K engagements

"Links: - ML for Beginners: - AI for Beginners: - NN Zero to Hero: - Paper implementations: - Made with ML: - Hands-on LLMs: - Advanced RAG techniques: - Agents for Beginners: - Agents towards production: - AI Engg. Hub: Note: This roadmap moves toward LLMs NLP and AI agents after Made With ML but dont overlook CV and RL. They're equally powerful paths for AI engineers to explore"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-17 12:33:01 UTC 215.4K followers, 6025 engagements

"1 Soft-label Distillation Generate token-level softmax probabilities over the entire corpus using: - A frozen pre-trained Teacher LLM - An untrained Student LLM Train the Student LLM to match the Teacher's probabilities. Check this out👇"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:39:02 UTC 215.4K followers, 2292 engagements

"Let's build a "Chat with your Code" RAG app using Qwen3-Coder:"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-24 17:06:34 UTC 215.4K followers, 68.2K engagements

"7 MCP projects for AI Engineers (with video tutorials):"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-03 13:02:46 UTC 215.4K followers, 327.4K engagements

"For years gradient boosting has been the go-to for tabular learning. TabM is a parameter-efficient ensemble that provides: - The speed of an MLP. - The accuracy of GBDT. The visual below explains how it works. Let's dive in"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-14 12:34:32 UTC 215.3K followers, 18.5K engagements

"Mixture of Experts (MoE) is a popular architecture that uses different "experts" to improve Transformer models. The visual below explains how they differ from Transformers. Let's dive in to learn more about MoE"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:20 UTC 215.4K followers, 9170 engagements

"MoEs have more parameters to load. However a fraction of them are activated since we only select some experts. This leads to faster inference. Mixtral 8x7B by @MistralAI is one famous LLM that is based on MoE. Here's the visual again that compares Transformers and MoE"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-21 12:30:52 UTC 215.4K followers, 6151 engagements

"Turn messy PDFs into clean LLM-ready data Dolphin revolutionizes document image parsing with its analyze-then-parse approach making complex documents accessible to AI applications. Handles text tables formulas & figures. XXX% open-source lightning-fast solution"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-09 12:50:39 UTC 215.4K followers, 120.8K engagements

"Similarly we can define the other specialists: Image Audio and Document Specialist Agents The same code is used which is shared at the end"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-23 13:20:41 UTC 215.4K followers, 1119 engagements

"Knowledge graphs are insanely good at giving agents human-like memory Today we're building an MCP-powered memory layer that can be shared across all your AI apps like Cursor Claude Desktop etc. It's built using a real-time knowledge graph. XXX% open-source and self-hosted"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-06-08 13:31:43 UTC 215.4K followers, 158.1K engagements

"I decided to put together all my MCP posts in a single PDF. It covers: - The fundamentals of MCP - Explanations with visuals and code - XX hands-on projects for AI engineers Download link in next tweet"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-06-27 12:58:34 UTC 215.4K followers, 212.9K engagements

"LLMs learn not only from raw text but also from other models. Googles Gemma X and X for example were distilled from the larger Gemini model. Today we cover the three most common knowledgedistillation methods. Let's dive in 🚀"
@akshay_pachaar Avatar @akshay_pachaar on X 2025-07-25 12:38:57 UTC 215.4K followers, 2844 engagements