Hazem Sayed

Building a Production-Ready RAG System for Persian Language Processing I’m excited to share insights from developing an advanced Retrieval-Augmented Generation system specifically engineered for Persian/Farsi document processing, leveraging cutting-edge ML infrastructure and architectural patterns. 🔬 Technical Architecture Highlights: Advanced RAG Implementation: ∙ Self-Corrective Agentic RAG with adaptive query rewriting and iterative refinement ∙ Hybrid search architecture combining dense embeddings with SPLADE sparse representations ∙ Multi-agent orchestration enabling context-aware document interaction and general conversation modes Production Infrastructure: ∙ NVIDIA Triton Inference Server deployment on H100 GPUs with vLLM backend ∙ Milvus vector database for high-performance similarity search ∙ PostgreSQL for structured data with MinIO object storage for scalable file management ∙ Model ensemble: Qwen2.5-72B-Instruct-AWQ, DeepSeek-OCR, and sentence-transformers ML Pipeline Innovations: ∙ Automated synthetic QA pair generation using LangGraph workflows ∙ Cross-encoder reranking (ms-marco-MiniLM-L6-v2) for precision optimization ∙ Dynamic prompt management system with A/B testing capabilities and versioning ∙ Fine-tuning pipeline with quality-scored synthetic datasets Engineering Principles: ∙ Clean Architecture with dependency injection and separation of concerns ∙ Docker containerization for reproducible deployments ∙ Comprehensive observability with metrics tracking and performance monitoring This project demonstrates how to architect enterprise-grade RAG systems that combine theoretical ML advancements with practical production considerations, particularly for low-resource languages. Open to discussing technical implementation details, architectural decisions, or challenges in deploying LLMs at scale. #MLEngineering #RAG #NLP #LLMs #VectorSearch #ProductionML #AI #DeepLearning #Persian #TritonInference #CleanArchitecture

1

AI agents are only as powerful as their memory. Redis has been named the #1 data storage choice for AI agents in 2025. The reason? Redis delivers what AI truly needs: - Real-time vector storage - Semantic caching - Long-term memory for LLMs At Bassirah, we see this as validation of why Redis is central to the next generation of AI-native data platforms powering everything from agentic RAG systems to enterprise decision engines. 🔗 Read more: https://lnkd.in/db47k6zG #AI #Redis #RAG #LLM #DataEngineering #AIAgents #ArtificialIntelligence

2

What does it take to succeed in data science? Strong coding and AI/ML knowledge are essential — but they’re just the beginning. Check out my talk at the next MLOps Denver meetup where I'll share five key lessons from over eight years in the field. From advocating for yourself to building cross-functional collaboration, this talk goes beyond the technical to explore what truly drives long-term success in a technical field. Hope to see you there! #DataScience #CareerGrowth #AI #MachineLearning #WomenInTech #ProfessionalDevelopment

13

1 Comment

AI is no longer just predicting — it’s planning, deciding, and acting. As we move into the era of agentic AI, the biggest question is no longer “How smart is the model?” It’s “How do we keep autonomous systems safe, reliable, and aligned with human intent?” That’s where AI guardrails come in. In my latest article, I explore: 1. What guardrails really mean for AI agents 2. Where they sit across the AI workflow (input → reasoning → action → monitoring) 3. How industries like finance, healthcare, manufacturing, and customer care use guardrails in practice 4. Why guardrails are not constraints—but enablers of trustworthy autonomy Powerful AI without boundaries is fragile. Well-designed guardrails are what make autonomy production-ready. If you’re building or deploying AI agents, this is no longer optional—it’s foundational. Link: https://lnkd.in/gvjeEGkg #AI #AgenticAI #AIGovernance #AIEngineering #ResponsibleAI #LLMOps #DataScience #LLM #GenAI #ChatGPT #Gemini

4

Excited to host our next Prompt by Design workshop at UT San Antonio with James B.! On Friday, March 20th (10:00 am–12:30 pm) at San Pedro I – Weston Conference Center, we will be running: Prompt by Design: Processing Real-World Multimodal Data with AI Vision-language models can analyze images, extract information, and answer complex questions, but getting reliable results requires thoughtful prompt design. In this hands-on session, we focus on practical, real-world multimodal tasks using modern vision-language models . We will cover: - Image classification - Caption generation - Multimodal question answering Practical techniques to improve reliability Through guided exercises, attendees will programmatically build working solutions for image-driven tasks. No prior coding experience is required . This workshop is sponsored by the Center for Data Science, USAA Student Success Center, and USAA Compute Hub . If you are interested in applied LLMs, multimodal AI, or building practical AI systems, this will be a very hands-on session. Register here: https://lnkd.in/dD5V7ntq Hope to see you there! #AI #LLMs #MultimodalAI #PromptEngineering #DataScience #UTSA UT San Antonio College of AI, Cyber and Computing James B. Fred Martin Debra Leal

59

3 Comments

New research from Databricks. It's about training enterprise search agents via RL. KARL introduces a multi-task RL approach where agents are trained across heterogeneous search behaviors, constraint-driven entity search, cross-document synthesis, and tabular reasoning. It generalizes substantially better than those optimized for any single benchmark. KARL is Pareto-optimal on both cost-quality and latency-quality trade-offs compared to Claude 4.6 and GPT 5.2. With sufficient test-time compute, it surpasses the strongest closed models while being more cost efficient.

396

8 Comments

Select AI Retrieval Augmented Generation now supports in-database ONNX-format transformers Last year, we announced Select AI Retrieval Augmented Generation (RAG) for Oracle Autonomous Database, which uses Oracle Database 23ai AI Vector Search and transformers hosted by your choice of supported AI providers. Cloud-based transformers (also known as embedding models) require the data to make a round trip to and from the transformer for creating the vector index, as well as for each prompt. We’re announcing general availability for in-database transformers with Select AI RAG that streamlines the process: It now eliminates the round trip by using transformers with the in-database ONNX Runtime. https://lnkd.in/ebUWJg_t #oracle #database #generativeai #oracleselectai #retrievalaugmentedgeneration #transformer #embeddingmodel

78

1 Comment

🎥 Recording Available: Deep Agents with LangGraph - Master Planning & Persistent Reasoning 🧠 Join Syed Hyder Zaidi, Senior Software Engineer at Data Science Dojo, for an in-depth session on building Deep Agents - the next evolution of AI systems that plan, remember, and reason across complex workflows. Learn how LangGraph makes it possible to go beyond simple LLMs and create stateful, production-ready AI agents with memory, collaboration, and persistence built in. 📅 Originally held on: Dec 10, 2025 | ⏰ 12:00 PM PT 👉 Watch the recording now: https://hubs.la/Q03Z1m0G0 What you’ll learn: - Why traditional agents fail on long-running, multi-step tasks - The 4 pillars of Deep Agents: planning, sub-agents, memory, and state - How LangGraph enables persistent reasoning and feedback loops - Building Deep Agents with observability, error recovery, and scaling - Real-world examples and a live walkthrough #LangGraph #DeepAgents #AgenticAI #AIWorkflows #LLMs #PersistentReasoning #AIArchitecture #FutureOfAI #DataScience

9