Senior AI Engineer

Job Title: Senior AI EngineerLocation: Urla, IzmirAbout Efsora:We are a fast-growing software and AI development company that partners with innovative enterprises and scaleups to deliver large, cutting-edge R&D projects. Our teams work as extensions of our clients' internal R&D, combining advanced technology expertise to build impactful solutions. We focus on augmenting our clients' R&D capabilities, accelerating innovation, and managing technical risks from early-stage prototyping to full-scale deployment.About the Role:We are seeking a highly experienced Senior AI Engineer with 5+ years of experience to lead the development of multi-modal systems. This role requires a balanced command of LLMs and Computer Vision, enabling you to build agentic architectures that can not only "reason" but also "see," and interpret the world. You will play a pivotal role in designing, developing, and deploying goal-oriented AI solutions. You will move beyond text-only interfaces to build agents that process visual inputs to automate complex problem-solving and drive autonomous operational efficiency.Responsibilities:Multi-Modal Architecture & Agents Lead the design and implementation of multi-modal agentic architectures that leverage LLMs as reasoning engines alongside specialized Computer Vision models.Develop robust orchestration layers that enable agents to ingest video streams or images, interact with external APIs, and execute complex actions based on multi-sensory synthesis.Design strategies for agents that utilize multi-step reasoning involving semantic understanding (text) and visual recognition (objects, OCR).Engineering & Optimization Design, develop, and maintain backend services and APIs that power multi-modal applications, ensuring scalability and performance for high-throughput media processing.Implement specialized techniques for prompt engineering and computer vision pipeline optimization to ensure accurate context management and model performance.Fine-tune LLMs and domain-specific Computer Vision models on custom datasets.Deployment & Evaluation Deploy and maintain sophisticated AI systems in production, ensuring reliability, cost optimization (e.g., managing token and GPU usage), and safety.Develop evaluation metrics for multi-modal agents, focusing on task completion rates, accuracy across modalities, efficiency, and robustness.Collaborate with product managers to translate complex requirements into practical, autonomous AI solutions and lead technical discussions on system architecture.What We're Looking For:Experience & Core SkillsProfessional Experience: 5+ years of professional experience in developing and deploying AI/ML models.Programming: Expert proficiency in Python and deep familiarity with libraries such as PyTorch, TensorFlow/Keras, and OpenCV.Problem-Solving: Exceptional ability to architect and debug complex, interdependent AI systems that span multiple modalities.Balanced Domain Expertise (LLM & Vision) Computer Vision Mastery: Deep experience with CNNs, Vision Transformers (ViTs), and standard tasks like Object Detection (YOLO, Faster R-CNN), Segmentation, and OCR.LLM Proficiency: Extensive experience with Large Language Models, including fine-tuning via Hugging Face Transformers and utilizing advanced APIs (OpenAI, Anthropic, Gemini).Multi-Modal Integration: Practical experience integrating structured outputs from Computer Vision models into LLM contexts to ground text generation in visual inputs.Agentic Design: Proven track record in designing agentic architectures, specifically involving planning loops, tool use, and memory management. Bonus Points If You Have: Voice AI & Audio: Experience with ASR (Automatic Speech Recognition) and TTS (Text-to-Speech) technologies (e.g., OpenAI Whisper, ElevenLabs) for building voice-enabled conversational agents.Orchestration Frameworks: Direct experience with agent orchestration frameworks (e.g., LangGraph, LlamaIndex, AutoGen).Video Analytics: Experience applying AI agents to real-time video feeds or CCTV data.MLOps & Cloud: Experience deploying complex, stateful systems on AWS, Azure, or GCP using Docker and Kubernetes.Model Context Protocol: Familiarity with MCP.Prototyping: Experience with Streamlit or Gradio for rapid multi-modal demos.