We are looking for a detail-oriented and technically curious  AI Trainer

 to support the development, training, and evaluation of LLM-based systems.

Project info:

 The project is an AI-powered virtual assistant for a telecom customer support. It uses a hybrid architecture combining Dialogflow CX, Generative AI, and advanced NLU with RAG tools to deliver accurate, contextual responses across voice and chat channels.

This is a hybrid role combining content expertise, linguistic precision, and a basic understanding of model behavior, particularly in the context of NLP and generative AI systems.

Key Responsibilities

• Design and optimize prompt templates for various use cases (e.g., summarization, classification, entity extraction, reasoning, question answering, dialogue).
• Evaluate model responses (using similarity comparison, criteria ranking, or pass/fail) and provide actionable feedback to fine-tune performance.
• Create high-quality input-output examples, instruction sets, and single-turn/multi-turn test cases for model behavior validation.
• Support adversarial testing and edge-case coverage to ensure the robustness of AI models.
• Analyze and troubleshoot model errors to identify failure patterns, inconsistencies, and hallucination risks.
• Perform post-deployment human evaluations to identify edge cases, model misalignments, and opportunities for improvement.
• Monitor live systems for intent recognition accuracy, fallback behavior, and GenAI output consistency; troubleshoot anomalies and implement fixes.
• Implement, assess and enhance the consistency and relevance of outputs from Retrieval-Augmented Generation (RAG) tools and Knowledge Bases/Datastores.
• Annotate, curate, and clean datasets for supervised and reinforcement learning purposes.
• Ensure all data handling complies with privacy, security, and ethical AI guidelines.
• Participate in the design and feasibility review of conversation flows, ensuring the AI system is well-equipped to handle real-world scenarios.
• Collaborate with engineering, QA, and product teams to ensure timely delivery of NLU and GenAI components as per the roadmap.

Requirements:

• Strong written English skills (C1+), with the ability to create clear, structured, and instruction-based content.
• Experience with prompt engineering and optimization for tasks such as summarization, classification, entity extraction, reasoning, Q&A, and dialogue.
• Understanding of LLM behavior and hands-on experience with models like OpenAI GPT, Anthropic Claude, and Google Gemini.
• Knowledge of RAG frameworks, vectorbases, optimized chunking techniques.
• Hands-on skills in data annotation and evaluation workflows, including ranking, scoring, and linguistic analysis.
• Familiarity with GenAI safety, including prompt injection mitigation and red teaming practices.
• Familiarity with conversational AI platforms such as Dialogflow, RASA, LUIS.ai, IBM Watson, Infobip, Amazon Lex.
• Proficiency in Python (pandas, numpy, spacy, nltk, scikit-learn, transformers, tensorflow, pytorch, matplotlib) for working with datasets and prototyping evaluations.
• Ability to work with structured data using SQL (PostgreSQL, MySQL, MongoDB) and tools like Databricks, Athena, Power BI, Looker etc.
• Basic familiarity with DevOps/testing tools (Docker, CI/CD, Postman).
• Use of collaborative tools such as Figma, Miro, Lucidchart for coordinating with design and product teams.