Tomas Baidal — Data Platform Engineer

Data Platform Engineer based in Berlin. I currently own MILES Mobility’s data platform end-to-end — ingestion, warehouse, transformation, BI, scheduling, observability, and governance — as the sole data engineer on a small team, shipping AWS infrastructure with Terraform, dbt, and observability built in.

Previously I scaled analytics infrastructure at Trade Republic Bank. I focus on governed, reproducible metrics and on using AI where it earns its place on the platform itself. Bachelor’s in Applied Data Science from UOC.

MILES Mobility / Data Analytics Engineer

Sep 2025 – Present, Berlin

Own the data platform end-to-end: ingestion, warehouse, transformation tooling, BI, scheduling, observability, and governance.
Rearchitected AWS and Snowflake infrastructure with Terraform, applying IaC best practices including least-privilege IAM; migrated legacy systems, deployed Airbyte OSS and custom connectors, and run platform services on EKS (Metabase, OpenMetadata, Cube, Temporal, Langfuse).
Architected an ML pipeline to train and infer a model that labels risky trips, using RabbitMQ over AMQPS, AWS Lambda, S3, and Terraform.
Built an internal FastAPI data API to orchestrate AI agentic workflows across the company.
Built the governed analytics layer: dbt models with tests, Cube as semantic layer, OpenMetadata for lineage and cataloguing, and Metabase for BI — so metrics are reproducible and auditable.
Orchestration with Temporal and GitLab CI/CD; Kafka for event-driven ingestion where stream processing fits the use case.
Applied AI selectively on the platform — Langfuse for LLM observability, automated triage, and internal agentic tools — where tasks are repeatable and verifiable.
Stood up MLOps for model training, experiment tracking, and production serving (SageMaker, MLflow, Langfuse).

Trade Republic Bank / Analytics Engineer, Operations & CX

Jan 2023 – Sep 2025, Berlin

Designed and maintained ELT pipelines on AWS and Airflow at scale across millions of customers — batch ingestion into Snowflake with dbt transformations and tested models.
Built operational datasets in dbt covering customer care KPIs, agent performance, and metrics surfaced through Looker and Metabase.
Led technical ownership of Zendesk: platform architecture, custom app development, and workflow automation via Zendesk Integration Services (ZIS).
Integrated third-party platforms via REST APIs; partnered with product engineering on event data and operational automation.

Trade Republic Bank / Platforms & Applications Specialist

Jan 2022 – Jan 2023, Berlin

Owned the Customer Service technology stack, including Zendesk administration, UltimateAI configuration, and API integrations.
Automated support workflows and built custom Zendesk apps to improve agent efficiency and data quality.
Acted as the technical bridge between CX and engineering, translating operational needs into scalable platform solutions.

Universitat Oberta de Catalunya / B.Sc. Applied Data Science

Degree covering data science, data engineering, machine learning, big data, and distributed systems. Excelled in Python, scripting, data warehousing, linear algebra, statistics, object-oriented programming, multivariate analysis, big data, and distributed systems.

Languages: Python, SQL, R, Java

Warehousing & OLAP: Snowflake, Redshift, PostgreSQL, MongoDB

Transformation & orchestration: dbt, Spark, Hadoop, Kafka, Airflow, Temporal, RabbitMQ (AMQPS), GitHub Actions

Cloud & IaC: AWS (EKS/Kubernetes, ECS, Fargate, EC2, IAM, S3, Lambda, Glue, CloudWatch, DMS, SNS, SQS), Terraform

DevOps: Docker, GitHub, GitLab CI/CD, Git, Terraform, FastAPI

BI, semantic layer & governance: Metabase, Cube, Looker, Superset, OpenMetadata, data contracts, lineage

AI / MLOps: SageMaker, Bedrock, Langfuse, MLflow

AWS Object Detection (MXNet / TensorFlow)

ML pipeline deployed on AWS

End-to-end airplane object-detection pipeline: EDA, augmentation, MXNet ResNet-50 training on SageMaker (benchmarked against TensorFlow), then real-time inference via Lambda, API Gateway, and a Dash web app.

AWS Image Classification (ResNet-50)

ML pipeline deployed on AWS

Dog-breed image classifier using a ResNet-50 base from torchvision, trained and deployed on AWS SageMaker — data prep, transfer learning, hyperparameter tuning, and endpoint inference.

ReDI School of Digital Integration / Data Analytics Instructor

Nov 2023 – Jun 2024, Berlin

Taught Python, Statistics, and Machine Learning to newcomers and locals as part of ReDI’s free tech education program — a non-profit focused on digital skills and job integration for underrepresented communities in Germany.

Spanish, Catalan (native), English (C1), German (C1)