Mid-Level Artificial Intelligence Engineer

Objective The objective of the Individual Contractor (IC) is to design, develop, and deploy production-grade agentic AI systems and LLM-powered solutions in support of the Arab Development Portal and the broader DSDSD innovation mandate. The IC will remain at the forefront of the rapidly evolving AI landscape, translating the latest research advances in large language models, multi-agent frameworks, and AI-assisted workflows into tangible organizational capabilities.

Background This position is located in the Decision-Support and Data Science Division (DSDSD). The Division is part of ESCWA's broader modernization and innovation efforts, providing advanced analytics and decision-support services within ESCWA, other UN entities, and Member States. Aligned with the UN 2.0 agenda and grounded in strategic foresight, DSDSD leverages data-driven insights, emerging technologies, and scenario-based planning to anticipate trends and proactively inform policymaking and operations. Its core functions span data integration, data quality assurance, advanced analytics and modeling, machine learning and AI solutions, real-time dashboards, performance reporting, and the design of specialized digital decision-support tools. Through these capabilities, DSDSD empowers evidence-based decision-making, fosters organizational efficiency, and catalyzes strategic innovation across the region. Work Assignment 1. Agentic AI System Design and Development • Architect and implement multi-agent systems using frameworks such as LangGraph, AutoGen, CrewAI, or equivalent, capable of orchestrating complex reasoning, retrieval, and action pipelines. • Design and deploy Retrieval-Augmented Generation (RAG) systems with advanced retrieval strategies (hybrid search, re-ranking, contextual compression) for knowledge-intensive tasks. • Implement tool-use and function-calling patterns enabling agents to interact with external APIs, databases, and computational tools. • Develop agent evaluation frameworks, including automated benchmarking, failure analysis, and iterative improvement loops. 2. LLM Integration and Optimization • Integrate and benchmark state-of-the-art LLMs (proprietary and open-source) across use cases including information extraction, summarization, classification, and question answering on Arabic and multilingual content. • Apply prompt engineering, structured output generation, and chain-of-thought techniques to maximize model reliability and task performance. • Explore and implement fine-tuning strategies (LoRA, QLoRA, full fine-tuning) where domain adaptation is required. • Monitor and optimize inference costs, latency, and throughput for production deployments. 3. AI Infrastructure and API Development • Design and implement RESTful and streaming APIs to expose AI capabilities to internal systems and the ADP platform. • Containerize AI services using Docker and manage deployments in line with organizational infrastructure standards. • Maintain documentation, versioning, and model cards for all deployed AI systems. 4. Research, Innovation, and Collaboration • Continuously monitor and evaluate emerging AI research, frameworks, and tooling to identify adoption opportunities aligned with DSDSD's mission. • Collaborate with data engineers, ML engineers, and domain experts to integrate AI capabilities into end-to-end data products. • Prepare technical reports, demonstrations, and presentations to communicate AI system design and findings to both technical and non-technical stakeholders.

A bachelor's degree in computer science, artificial intelligence, data science, or a related field is required. A master's degree in computer science, artificial intelligence, data science, or a related field is desirable. All candidates must submit a copy of the required educational degree. A minimum of 5 years of professional experience in AI/ML engineering or software engineering with a demonstrated AI focus is required. Hands-on experience designing and deploying agentic systems using multi-agent frameworks (LangGraph, AutoGen, CrewAI, or equivalent) is required. Proficiency in Python and experience with LLM APIs (OpenAI, Anthropic, Mistral, or open-source alternatives via Ollama or HuggingFace) is required. Demonstrated knowledge of the current LLM and AI landscape, including frontier model capabilities, benchmarks, and limitations, is required. Experience with RAG architectures, vector databases (FAISS, Weaviate, Qdrant, or equivalent), and semantic search is desirable. Familiarity with model fine-tuning, RLHF, and evaluation methodologies is desirable. Knowledge of Arabic NLP and multilingual model handling is desirable.

English and French are the working languages of the United Nations Secretariat; and Arabic is a working language of ESCWA. For this position, fluency in English s required. Note: “Fluency” equals a rating of ‘fluent’ in all four areas (speak, read, write, and understand) and “Knowledge of” equals a rating of ‘confident’ in two of the four areas.

Not available.