AI Agent Developer (LLM Specialist) Role Overview The AI Agent Developer is responsible for designing, building, and deploying intelligent autonomous or semi-autonomous agents powered by Large Language Models (LLMs). This role focuses on enabling complex workflows, reasoning capabilities, system integration, and secure data interaction across enterprise environments. Key Responsibilities 🔹 Agent Design & Development - Architect and implement AI-powered agents using LLMs (e.g., GPT, Claude, Llama, Mistral, Gemini). - Develop prompt workflows, retrieval-augmented generation (RAG), and multi-step reasoning. - Configure agent behaviors for decision-making, task execution, and contextual understanding. - Implement function calling, tool use, or API integration to extend agent capabilities. 🔹 System Integration - Connect agents to internal systems, databases, knowledge bases, and APIs. - Use frameworks such as LangChain, LlamaIndex, Haystack, Semantic Kernel, or custom pipelines. - Implement connectors for CRMs, ERPs, ticketing platforms, or document repositories. 🔹 Data & Knowledge Management - Create structured and unstructured knowledge retrieval mechanisms. - Build and manage vector databases (Pinecone, Weaviate, Chroma, Milvus, Azure AI Search). - Implement embeddings, chunking strategies, metadata tagging, and query optimization. 🔹 Workflow Automation - Design agent orchestration pipelines to handle multi-agent collaboration. - Enable agents to automate research, reporting, email drafting, document processing, and customer interactions. - Integrate with workflow tools like Zapier, Airflow, UiPath, or Power Automate. 🔹 Model Optimization & Evaluation - Fine-tune or customize LLMs using organization-specific data. - Evaluate model performance for accuracy, latency, cost, hallucinations, and compliance. - Apply guardrails, prompt validation, and reinforcement techniques. 🔹 Security, Ethics & Compliance - Enforce data privacy and governance standards (GDPR, HIPAA, SOC2, ISO, CCPA). - Implement role-based access and secure API token handling. - Apply content filtering, moderation, and usage monitoring. 🔹 Collaboration & Documentation - Work with cross-functional teams (engineering, design, product, legal, security). - Maintain technical documentation, agent configuration logs, and deployment playbooks. - Train internal teams on capabilities and responsible use.

• Adapt LLMs to diverse healthcare use-cases using techniques such as Sparse Fine-Tuning (SFT), Prompt Engineering Fine-Tuning (PEFT), Direct Parameter Optimization (DPO), and Proximal Policy Optimization (PPO). • Optimize LLMs for Retriever-Augmented Generation (RAG) to enhance decision-making and information retrieval capabilities. • Collect, clean, and refine healthcare datasets for training LLMs to ensure high-quality data provisioning. • Convert models into various formats suitable for production environments, ensuring their readiness for real-world application.

• Σχεδιασμός και ανάπτυξη λύσεων NLP με χρήση Large Language Models (π.χ. OpenAI, HuggingFace Transformers, LLaMA, Mistral κ.ά.). • Fine-tuning και βελτιστοποίηση προεκπαιδευμένων μοντέλων σε εξειδικευμένα datasets. • Ανάπτυξη pipelines για data preprocessing, training, evaluation και deployment σε cloud/edge περιβάλλοντα. • Συμμετοχή σε έργα R&D που ενσωματώνουν AI σε επιχειρησιακές και ερευνητικές πλατφόρμες. • Συνεργασία με Software Engineers, UI/UX Designers και Systems Architects για την ολοκληρωμένη υλοποίηση λύσεων. • Συμβολή στη συγγραφή τεχνικών παραδοτέων και τεκμηρίωσης AI modules. • Παρακολούθηση εξελίξεων στον χώρο του NLP & LLMs και εισαγωγή νέων μεθοδολογιών.

Please submit your resume in English and indicate your level of English. At Toloka, we connect smart, curious people from around the world with freelance online tasks that train and improve artificial intelligence. What we do The Toloka Annotators connects individuals with Generative AI projects from leading tech innovators. Our mission is to unlock the full potential of AI by involving real people from around the world in the development process. About the Role Annotation is what helps AI make sense of the world. As an annotator, you may be invited to take part in online projects such as rating AI-generated content, evaluating factual accuracy, or comparing responses - when projects are available. Responsibilities: - Carefully review provided data (text, images, or videos) - Label or classify content based on project guidelines - Identify and flag factually incorrect, sensitive, inappropriate, or unclear material Important note: This is project-based work. Tasks are available only when projects are active. You may be invited to one or more projects depending on your profile and current opportunities. Each project has its own compensation level based on scope and expertise required. On this project, AI trainers earn up to $23 per hour equivalent.