Full Stack Engineer

Fully Remote in Spain or Poland

We are working with a leading online scheduling platform designed to simplify the process of coordinating meetings and events. Founded over 18 years ago, it helps individuals and teams avoid the "back-and-forth" of email scheduling by allowing users to propose multiple time slots and let participants vote on their availability.

Responsibilities of the role:

• Architect Production AI Systems: Design reliable, production-ready AI systems, selecting optimal tools for robust real-world performance.
• Curate Data & Feature Stores: Prepare high-quality datasets and maintain feature stores to ensure data consistency for training and inference.
• Build Scalable ML Pipelines: Develop end-to-end data and ML pipelines using Airflow and dbt for seamless ingestion, deployment, and monitoring.
• Design & Deploy Models: Prototype and train diverse neural architectures, including LLMs, with a focus on reproducibility and performance.
• Implement Advanced Retrieval (RAG): Design Graph RAG and hybrid retrieval systems, including graph construction and entity linking.
• Enable Edge Intelligence: Optimize and quantize large models for efficient on-device and edge processing.

Requirements of the role:

• Experience delivering complete AI components—from planning and modeling to deployment, monitoring, and iteration.
• Strong Python skills and deep familiarity with ML frameworks such as Scikit-Learn, TensorFlow, PyTorch, and Hugging Face. You’re comfortable designing, evaluating, and prototyping diverse model types.
• Hands-on experience with MLOps tools (e.g., MLflow, ZenML), dbt modeling, and working with cloud data warehouses or data lakes.
• Experience building and scheduling pipelines in Airflow. Familiarity with modern data stacks such as Kafka, Spark, and cloud warehouses (BigQuery, Redshift, Snowflake). Ability to define event-level tracking schemas for reliable analytics.
• Strong understanding of model behavior and evaluation. Experience developing frameworks for assessing model quality, reliability, hallucination detection, prompt regression, safety scoring, or multi-hop reasoning. Familiarity with RAG, graph-based retrieval, and prompt design.
• A focus on shipping systems that are robust, explainable, and usable by others.