Senior AI Engineer – Digital Health (Ethiopia)

About the role

Last Mile Health is seeking a Senior AI Engineer to lead the design, development, and deployment of advanced AI‑powered digital health solutions for Health Extension Workers in Ethiopia. The role focuses on multilingual AI, multimodal interactions, and offline‑first capabilities to support low‑connectivity environments.

Key responsibilities

• Design and enhance the HEP Assist AI platform, adding support for Amharic and other local languages.
• Develop prompt‑engineering workflows, fine‑tune and evaluate models for healthcare use cases.
• Build multimodal AI features such as voice‑to‑voice chat, speech recognition, and audio/video processing.
• Implement offline‑first AI functionality and Retrieval‑Augmented Generation (RAG) and Graph RAG pipelines.
• Create diagnostic AI agents and agentic workflows to aid health‑decision making.
• Develop secure backend APIs for data ingestion, processing, and exchange.
• Conduct model benchmarking, hallucination testing, and LLM‑Judge assessments.
• Integrate AI systems with telephony, messaging, mobile, web, IVR, and voice‑based services.
• Design scalable infrastructure for model serving, vector databases, document pipelines, and real‑time AI interactions.
• Collaborate with cross‑functional teams to prioritize the AI product roadmap and deliver high‑quality releases.

Required profile

• Senior‑level experience in AI/ML engineering, preferably in health or low‑resource settings.
• Located in Ethiopia with unsponsored work authorization.
• Strong ability to work onsite in Addis Ababa and collaborate with Ministry of Health and other stakeholders.

Required skills

• Multilingual AI system development (Amharic and other Ethiopian languages)
• Prompt engineering and model fine‑tuning
• Vector Retrieval‑Augmented Generation (RAG) and Graph RAG pipelines
• Agentic AI and LLM deployment in on‑premise environments
• DevOps automation, CI/CD pipelines
• Secure backend API development
• Speech recognition, voice‑to‑voice chat, audio and video processing
• Offline‑first AI design for low‑bandwidth environments
• Model benchmarking, hallucination testing, LLM‑Judge assessments
• Integration with telephony, messaging, mobile, web platforms, IVR, and voice services
• Scalable infrastructure for model serving, vector databases, document pipelines, real‑time AI interactions