نبذة عني
AI Engineer specializing in agentic AI systems, retrieval-augmented generation, and the deployment of open-source LLMs on production GPU infrastructure. Builds full-stack, agentic applications optimized for low token con…
AI Engineer specializing in agentic AI systems, retrieval-augmented generation, and the deployment of open-source LLMs on production GPU infrastructure. Builds full-stack, agentic applications optimized for low token consumption without sacrificing accuracy, and designs agentic RAG that feels natural to interact with. Experienced across NLP, document intelligence, and computer vision, with a focus on translating modern AI techniques into reliable, production-ready products.
الخبرة
AI Engineer
<p>Cut trade finance document review from 1 week to 3 minutes by designing and delivering a production RAG system for a major bank in Kuwait, handling letters of credit and guarantees across 100+ page documents.</p><p> Improved retrieval accuracy by 80% and eliminated OCR errors entirely by retrieving over document layout and visual structure rather than extracted text.</p><p> Reduced inference time from 20 minutes to 40 seconds (30x) through GPU efficiency work and token caching, cutting production latency and serving cost.</p><p> Built the system to sustain 20 concurrent users per instance with no measurable slowdown.</p><p> Owned the majority of the build within a 3-engineer AI team, and supported the system after go-live through monitoring, debugging, and iteration.</p>
AI Engineer
<p>Deployed hundreds of open-source models across multiple Linux GPU servers with varying hardware configurations, handling provisioning, serving, and scaling.</p><p> Built RAG applications with reasoning capabilities on top of deployed open models, including the embedding and retrieval pipelines behind them.</p>
AI Engineer Intern
<p>Developed text classification models for banking institutions.</p><p> Built LangChain applications leveraging OpenAI GPT models.</p>
المشاريع
Streamlind
<p>Developed a platform that allows users to build visually stunning, functional workflows using only plain English.</p>