Mid-level Data Science Analyst – AI

This is a fully remote position, open to applicants in Brazil.

📋 Description

• Engage in the training and assessment of models and prompts, focusing on fine-tuning, optimization of chunking and tokenization, bias identification, adversarial prompt evaluation, and the selection of embeddings to support RAG.

• Train and assess generative AI models, including large language models (LLMs) and multimodal architectures.

• Create and refine prompts, incorporating adversarial testing and risk management strategies.

• Conduct fine-tuning and hyperparameter optimization to enhance model performance.

• Apply chunking and tokenization techniques to improve efficiency and quality.

• Identify and address biases in datasets and models to ensure adherence to ethical standards.

• Choose and validate embeddings for applications involving Retrieval-Augmented Generation (RAG).

• Track quality, robustness, and safety metrics for models.

• Collaborate with engineers and architects as a subject-matter expert to securely embed models into production workflows.

⛳️ Requirements

• Expertise in Generative AI Models: LLMs, diffusion models, and Transformer architectures.

• Proficiency in Fine-tuning and Prompt Engineering: advanced methods, including RLHF (Reinforcement Learning from Human Feedback).

• Knowledge of Tokenization and Chunking: strategies aimed at optimizing context.

• Experience with Embeddings and RAG: selection, assessment, and integration with retrieval/search systems.

• Understanding of Bias & Fairness: methods for detecting and alleviating biases.

• Security and Robustness skills: adversarial testing and prevention of jailbreaks.

• Familiarity with Tools and Frameworks: ADK, CrewAI, Agno, LangChain, LangGraph.

🏝️ Benefits

• Remote work model

• CLT employment (Brazilian statutory employment contract)

• Flexible working hours