AI Engineer (Arabic NLP & Predictive Models , LLM)

About Kurasty

Kurasty is a Saudi startup building the next generation of AI-powered procurement tools. We develop lightweight, high-impact add-ons and plugins that integrate with leading ERP systems (SAP, Oracle, Microsoft Dynamics 365) or run as standalone SaaS tools. Our mission is to empower government and enterprise procurement teams with intelligent, automated tools that drastically cut time, reduce errors, and ensure compliance.

About the Role

We’re hiring a skilled AI Engineer to power the intelligence behind our procurement tools. You’ll work on everything from Arabic NLP to document classification, predictive analytics, and optimizing LLMs. Your work will directly affect how government entities create, review, and analyze complex documents.

You will collaborate closely with the back-end team (who will orchestrate your models) and the DevOps team (who will deploy your containers in production).

Responsibilities

Build and fine-tune Arabic NLP models / LLMs for tasks like content generation, classification, summarization, and named entity recognition

Build accurate document parsing and data extraction pipelines from different file types (DOCX, PDF, XLSX)

Use prompt engineering and embedding models for retrieval-augmented generation (RAG)

Leverage LLMs to provide nuanced and accurate information (Chunking, Rate limit optimization, prompt engineering)

Package AI models into deployable Python containers

Collaborate with cross-functional teams to ensure AI outputs are fast, reliable, and human-readable

Leverage Kurasty’s proprietary dataset of RFPs and government documents to improve model accuracy

Must-Have Qualifications

1–3 years of experience in machine learning, NLP, or applied AI

Hands-on experience with Arabic NLP, including tokenizer customization and right-to-left handling

Proficient in Python and ML libraries like HuggingFace Transformers, Scikit-learn, SpaCy, FastAPI, Pydantic, Poetry, LangChain (LangGraph), etc.

Experience building, training, and evaluating custom or fine-tuned models

Strong understanding of prompt engineering and LLM integration

Fluent in both Arabic and English

Based in Riyadh or willing to relocate (on-site only)

Preferred Qualifications

Experience working with procurement, government documents, or legal texts

Familiarity with vector databases, embeddings, and search optimization

Background in deploying models to production environments

Familiarity with OCR tools like Tesseract, Textract, or Google Vision API

Exposure to ERP systems and procurement domain is a plus

Show more Show less