Update README.md

README.md

@@ -38,83 +38,36 @@ thumbnail: >-
      alt="Navid Banner" 
      style="width:100%; height:200px; object-fit:cover;"/>
 
-_**Building Arabic-first AI with Navid**_
+_**We are a team that seeks to connect our people's past with the world's future.**_
 
-At **Navid**, we design, train, and deploy Arabic-first and multilingual AI systems. From curated data pipelines and synthetic data generation to custom LLM training, evaluation, and production-grade RAG/agentic apps, we help teams in KSA and beyond ship robust AI—on-prem or in the cloud.
+At **Navid**, we build Arabic-first and multilingual GenAI solutions across domains, laser-focused on delivering real value for people and organizations.
 
-## Build and Scale GenAI with Navid
+## What We Do
 
-Navid offers an end-to-end stack spanning data, models, and applications. We specialize in:
+- **Data pipelines & governance** — automated collection, filtering, and curation  
+- **Synthetic data & augmentation** — domain-balanced recipes for Arabic tasks  
+- **Training & fine-tuning** — SFT, DPO, PPO, RLHF/RLAIF  
+- **Custom LLMs & evals** — Arabic/multilingual models with robust evaluation  
+- **Advanced RAG** — secure retrievers, re-rankers, hybrid search & guardrails  
+- **Agentic workflows** — LangGraph/LangChain multi-agent apps with memory/tools  
+- **Deployment** — on-prem, VPC, or managed GPU with full observability  
 
-- **Data pipelines & governance** — automated collection, cleansing, deduping, safety filtering, and curation.
-- **Synthetic data & augmentation** — domain-balanced recipe design for low-resource Arabic tasks.
-- **Training & fine-tuning** — SFT, DPO/PPO, curriculum & multi-stage training on open and custom architectures.
-- **Custom LLMs & evals** — Arabic/multilingual capabilities, eval harnesses, human+AI QA, and enterprise red-teaming.
-- **Advanced RAG** — document intelligence, retriever/re-ranker selection, vector/hybrid search, caching & guardrails.
-- **Agentic workflows** — LangGraph/LangChain-based multi-agent systems with tools, memory, and supervision.
-- **Secure deployment** — on-prem, VPC, or managed GPU (e.g., vLLM, Triton) with observability (Prometheus/Grafana/Jaeger).
+🚀 **Highlight:** *Yehia* — one of the top models for its size trained by Navid.  
 
-Whether you need domain-tuned assistants, multilingual search, or end-to-end AI transformation aligned with **Vision 2030**, Navid provides production-grade solutions tailored for enterprise scale and policy compliance.
+## Building Agents
 
-Learn more about our capabilities:
+- **Planner–Worker patterns** for complex tasks  
+- **Structured tool-use** with safety constraints  
+- **Grounded reasoning** with hallucination grading  
+- **Replayable traces & guardrails** for compliance  
 
-* Arabic-first **LLM Training & Fine-tuning**
-* **On-prem & VPC RAG** for regulated data
-* **Evaluation & Benchmarks** for Arabic/multilingual tasks
-* **Agentic Applications** for workflows & decision support
+Use cases: knowledge copilots, policy Q&A, analytics, code assistants, and enterprise automation.  
 
-> _Ask us about Navid Hub, our foundation for dataset/versioning, experiment tracking, and deployment automation._
+## Open Source
 
-## Building AI Agents with Navid
+We contribute to the Arabic AI ecosystem with:  
 
-We design multi-agent systems for retrieval, planning, reasoning, and actions across enterprise tools. Our approach combines:
-
-- **Structured tool-use** with typed actions and safety constraints
-- **Planner–Worker patterns** using LangGraph
-- **Grounded answering** via graded hallucination control
-- **Replayable traces & guardrails** for compliance and audit
-
-Use cases: Arabic knowledge copilots, policy Q&A, financial analytics, code assistants, and L4/L5 support automation.
-
-* Get started with **Agentic Templates** (LangGraph/LangChain)
-* Production **Evaluation Harness** for agents (trace-based)
-
-## Navid & Open Source
-
-We contribute to the Arabic AI ecosystem: datasets, re-rankers for RAG, evaluation recipes, and MLOps utilities. Highlights include:
-
-- **Arabic/Multilingual Reranker Leaderboard**
-- **RAG Evaluation Recipes** (faithfulness, context precision/recall, semantic similarity)
-- **Data loaders & cleaners** for Arabic PDFs and OCR post-processing
-
-* Explore sample repos and templates (fill in links below):
-  * GitHub: `<your_org_url_here>`
-  * Docs: `<your_docs_url_here>`
-
-## Connect, Learn, and Grow with Navid
-
-<div class="grid lg:grid-cols-2 gap-x-4 gap-y-7 p2">
-  <div class="col-span-1 p2">
-    <p>Navid Community</p>
-    <ul class="social">
-      <li><a href="#">Navid on GitHub</a></li>
-      <li><a href="#">Navid on LinkedIn</a></li>
-      <li><a href="#">Open Source at Navid</a></li>
-    </ul>
-  </div>
-  <div class="col-span-1 p2">
-    <p>Research & Datasets</p>
-    <ul class="social">
-      <li><a href="#">Navid Research on GitHub</a></li>
-      <li><a href="#">Code & Datasets</a></li>
-    </ul>
-  </div>
-  <div class="col-span-1 p2">
-    <p>Navid Developers</p>
-    <ul class="social">
-      <li><a href="#">Navid Engineering Blog</a></li>
-      <li><a href="#">Generative AI at Navid</a></li>
-      <li><a href="#">Navid Learning Hub</a></li>
-    </ul>
-  </div>
-</div>
+- **First Arabic RAG Leaderboard** — open-sourced for the community  
+- **Arabic/Multilingual Reranker Benchmarks**  
+- **Evaluation recipes** for faithfulness, recall, precision, and similarity  
+- **Data loaders/cleaners** for Arabic PDFs & OCR