Hi there, I’m Asif Mehmood 👋

AI Engineer | Machine Learning Specialist | LLM & RAG Expert | Agent Workflow Architect

• 🔭 I design and build end-to-end LLM applications, RAG pipelines, and ML solutions that power intelligent systems for enterprise workflows.
• 🌱 I build and optimize agentic workflows, prompt strategies, retrieval pipelines, and fine-tuning processes for production-ready GenAI systems.
• 👯 I love collaborating with other AI enthusiasts, researchers, and product builders.
• 🧠 Past projects include medical image segmentation, business-specific chatbots with RAG, document and data extraction pipelines, AI agents, OCR systems, and NLP automation tools.
• 🥅 2025 Goals: Contribute to open-source GenAI tools, grow my network, and help shape the future of applied AI.
• ⚡ Fun fact: I'm fascinated by AGI theory and enjoy translating complex AI concepts into simple stories that anyone can understand.

---

🌐 Connect With Me

---

🚀 Featured Projects

• 🔍 RAG-as-a-Service Platform – Modular retrieval-augmented generation system with multi-vector support.
• 🤖 Conversational AI with Tools & Memory – LangChain-powered chatbot with custom tools and state.
• 🧾 OCR-Powered Form Understanding – Document parsing + classification + summary pipeline.
• 🧠 Fine-Tuned LLM for Niche Domain – QLoRA + PEFT on domain-specific corpus.

---

🧠 Open Source Contributions

I've contributed to the LangChain and LangChain Community repositories, focusing on improvements to tools, utilities, and core features.

🔗 Notable Merged Pull Requests

• langchain-ai/langchain-community#83
  🔢 Added AWS Bedrock Claude 4 series model cost tracking
  Enhanced bedrock_anthropic_callback.py to include token pricing data for the following Claude 4 models:

  • anthropic.claude-sonnet-4-20250514-v1:0
  • anthropic.claude-opus-4-20250514-v1:0
    Pricing based on AWS Bedrock documentation.
    Reference: AWS Bedrock Pricing
    Related Issue: #82

• langchain-ai/langchain-community#50
  🔍 Added support for additional_search_client_options in Azure Search (async)
  Extended the Azure Search async vectorstore client to accept additional_search_client_options, aligning its behavior with the sync client.
  Previously, the async client ignored this parameter and defaulted to an empty dictionary.
  Related Issue: #42

• langchain-ai/langchain#31177
  📚 Fixed outdated import paths and corrected documentation typo
  Updated the import path of DoctranPropertyExtractor from document_loaders to document_transformers to match recent package refactoring.
  Also fixed a minor typo across docs for clarity and correctness.

• langchain-ai/langchain#31107
  ⚙️ Replaced deprecated .dict() calls with .model_dump() for Pydantic v2 compatibility
  Updated multiple modules to replace deprecated .dict() usage with .model_dump() to align with Pydantic v2+ guidelines and suppress PydanticDeprecatedSince20 warnings.
  This change prepares the codebase for Pydantic v3 with no impact on functional logic.
  Related Issue: #31103

• langchain-ai/langchain-community#27
  💰 Added token usage & cost tracking for OpenAI embedding models in OpenAICallbackHandler
  Enhanced get_openai_callback() to include embedding model calls in usage and cost reporting, improving observability and enabling complete cost tracking across all OpenAI model types.
  This change aligns LangChain's callback utilities with real-world production monitoring needs.
  Related Issue: #25888

🛠️ Ongoing Contributions

• Actively contributing additional improvements to LangChain and related projects.

---

✍️ Writing & Thought Leadership

I'm passionate about making AI accessible and understandable for everyone — from engineers to business leaders. I regularly write about real-world applications of AI, LLMs, RAG, and agentic workflows to share insights, simplify complex ideas, and spark meaningful conversations.

📝 Featured AI Articles

• 🔍 RAG Explained Simply for Non-Techs

Explore my work on Medium.

---

🧠 Tech Stack & Tools