• Analyse business requirements for the desired output format and the logic the model must implement. • Prepare datasets based on example texts: cleaning, annotation, creating training/validation splits. • Train and fine-tune LLMs for specific use cases: • configure training parameters; • experiment with prompts, system instructions, input/output formats. • Evaluate model quality: • design and track metrics; • create test scenarios and A/B experiments; • ensure output format consistency and stability. • Deploy models to Google Cloud (for example via Vertex AI, Cloud Run, Kubernetes, etc.). • Develop services and APIs (REST/gRPC) that expose the model to other systems. • Build automations and integrations that call the model: • background jobs, queues, event-driven triggers; • integration with internal services and databases. • Implement MLOps pipelines: • automate training / retraining workflows; • version models and datasets; • monitor model performance and quality in production. • Document models, pipelines, APIs, and architectural decisions.