AI Engineer - Agentic and Multimodal Solutions

About the role

As an AI Engineer at Tarjama, you will support the design, development and deployment of agentic and multimodal AI solutions based on language, vision and speech technologies. You will help build scalable, production-ready AI products by contributing to model integration, evaluation, optimization and data pipeline preparation, working closely with senior engineers.

Key responsibilities

• Design and implement AI agentic workflows that orchestrate language, vision and speech models for multi-step tasks, and build tool-using agents with decision logic, memory strategies and tool calling.
• Develop text, vision and speech systems: LLM pipelines, embeddings, rerankers, RAG, OCR-based document understanding, STT, TTS and conversational voice interfaces, plus multilingual translation pipelines.
• Own specific AI components from design through deployment, deploy and monitor systems in production, and work with engineering teams on scalable, maintainable integrations.
• Define evaluation metrics such as accuracy, latency and robustness, benchmark models, run experiments and A/B tests, and identify failure modes.
• Design and maintain data pipelines for text, document and audio data, ensure data quality, and analyze model outputs and user interactions.
• Collaborate with product managers, designers and engineers, partner with senior engineers on system design, and support integration, testing and deployment.
• Maintain documentation for pipelines, agent workflows and evaluations, share knowledge through code reviews and demos, and follow AI development best practices.