Role Description Na Deloitte, buscamos pessoas que queiram gerar impactos positivos todos os dias. Impulsionamos talentos para que possam se desenvolver em um ambiente colaborativo, com times diversos e que tragam energia, empoderamento, interação e conexões. O nosso crescimento é exponencial porque os talentos que recrutamos possuem os nossos valores em suas essências. Liderar o caminho, fomentar a inclusão, colaborar para mensurar impactos, servir com integridade e cuidar uns dos outros são pontos essenciais e inegociáveis para fortalecer ainda mais o nosso propósito. Você terá a oportunidade de evoluir sua carreira na maior empresa de serviços profissionais do mundo como Cientista de Dados Visão Computacional Pleno no time de desenvolvimento, aqui você terá a oportunidade de ser o protagonista de sua carreira, e de gerar impactos e transformações positivas em seu redor. Na sua rotina, você irá: - Desenvolver aplicações e pipelines de processamento em Python. - Treinar, ajustar e implantar modelos de deep learning em ambientes produtivos. - Integrar modelos generativos a pipelines de dados e sistemas de negócio. - Otimizar modelos para dispositivos de borda, utilizando ONNX, TensorRT e OpenVINO. - Projetar e manter soluções com bancos de dados relacionais, NoSQL e vetoriais (como Chroma e Faiss). - Implementar e acompanhar monitoramento e observabilidade com ferramentas como Prometheus, Grafana e CloudWatch. - Containerizar aplicações e serviços utilizando Docker. - Atuar em ambientes de nuvem AWS, apoiando desenvolvimento e operação das soluções. - Utilizar Git para versionamento de código e aplicar práticas de CI/CD. Qualifications - Ensino Superior Completo em cursos de TI; - Lógica de programação sólida e boas práticas de engenharia de software; - Linguagem de programação Python para desenvolvimento e pipelines de processamento; - Capacidade de treinar, ajustar e implantar modelos de deep learning; - Conhecimento com modelos generativos e integração desses modelos em pipelines; - Otimização de modelos para dispositivos de borda (ONNX, TensorRT, OpenVINO); - Experiência com bancos de dados (Relacional, NoSQL, Vetoriais: Chroma, Faiss, etc); - Familiaridade com ferramentas de monitoramento (Prometheus, Grafana, Cloudwatch); - Conhecimento com docker; - Conhecimento em ambientes de nuvem (AWS); - Conhecimento em versionamento com Git e práticas de CI/CD; Requirements - Desejável: Experiência com processamento de imagem e bibliotecas como OpenCV e scikit-image; - Experiência com modelos clássicos de machine learning e a biblioteca scikit-learn; - Experiência com modelos de deep learning e as bibliotecas PyTorch e/ou Tensorflow; - Experiência com modelos populares de detecção de objetos, como Yolo e SSD; - Experiência com message brokers, como: Apache Kafka, RabbitMQ ou Amazon Kinesis; Company Description Faz toda diferença contar com quem pode fazer dos sonhos, realidade. Escolha seu impacto!

• Partner directly with business stakeholders to identify opportunities where data and machine learning can improve decisions, efficiency, or outcomes • Design experiments and hypotheses that can be validated quickly using available data and pragmatic modeling approaches • Own models end-to-end—from data preparation and feature engineering through deployment, monitoring, and iteration based on real-world results • Deploy ML models into production using AWS-native tooling and integrate them into operational workflows and downstream systems • Implement ML training and inference pipelines on Amazon SageMaker, including pipelines, endpoints, model registry, and monitoring • Monitor model performance (accuracy, drift, stability, business KPIs) and iterate based on real-world impact • Build and operate data ingestion and transformation pipelines across batch and event-driven workloads using AWS Glue, zero‑ETL integrations, Step Functions, EventBridge, and related services • Collaborate closely with IT, Security, and Platform Engineering teams to align with enterprise security, compliance, and operational standards • Use infrastructure as code (Terraform, CDK, or CloudFormation) to create repeatable, scalable environments • Own and operate S3-based data lake infrastructure, including Iceberg table formats, AWS Glue Data Catalog, and AWS Lake Formation • Implement and enforce data zone architecture (e.g., raw, curated, and consumption zones) to support governed data access and lifecycle management • Define and apply data access controls using Lake Formation permissions and IAM-aligned policies • Establish and maintain data governance practices, including schema management, schema evolution, and lineage tracking • Ensure data assets are discoverable, auditable, and secure through cataloging, metadata management, and access controls • Build end-to-end observability using CloudWatch, Datadog, pipeline SLAs, data quality checks, and model drift detection • Establish operational runbooks and support procedures for governed data and ML platforms • Apply cost-aware design when selecting data processing, training, and inference approaches • Optimize Glue, SageMaker, and storage usage to deliver value efficiently at scale • Continuously improve platform reliability, scalability, and cost efficiency as data and ML workloads grow

About Hypha Metrics Hypha Metrics is redefining how media data is measured, connected, and understood. Our platform provides the foundational infrastructure that powers audience insights for the modern media ecosystem. We help clients make smarter decisions by transforming fragmented data into unified, actionable intelligence. Role summary Design and deliver rigorous, scalable measurement solutions that uncover true media impact and drive media investment decisions. This role combines causal inference, experimentation, statistical modeling, and production data engineering to measure incrementality, optimize media mix and inform campaign strategy across digital and offline channels. Key responsibilities - Design and analyze randomized and quasi-experimental tests (holdouts, geo-tests, RCTs) to measure advertising incrementality and lift. - Build and maintain causal models (difference-in-differences, synthetic controls, hierarchical Bayesian, uplift modeling) and marketing mix models (MMM) for multi-channel attribution. - Develop and productionize scalable end-to-end pipelines for event-level ad exposure, conversions, and offline-sales ingestion (ETL/ELT, validation, monitoring). - Work with data engineering to keep measurement datasets clean, deduplicated (identity resolution), and privacy-compliant. - Own feature engineering, model training, validation, and deployment in Python/R and cloud environments (BigQuery, Snowflake, Dataproc/AWS/GCP). - Produce clear, actionable dashboards and executive-ready insights for product, media, and client teams; present findings to stakeholders. - Implement automated reporting and CI/CD for model retraining and performance monitoring; establish measurement governance and documentation. - Stay current on the ad-tech/measurement ecosystem (ATtribution frameworks, walled gardens, ID solutions) and recommend measurement strategy changes. Required qualifications - 3+ years experience building statistical or ML models in ad-tech, marketing analytics, agency measurement, or a related data science role. - Strong statistics/causal inference fundamentals: experimental design, hypothesis testing, regression, hierarchical models. - Proficient in Python (pandas, scikit-learn, PyMC/Stan or equivalent) and SQL; experience with R is a plus. - Hands-on experience with event-level ad/exposure and conversion data, logs from ad servers/DSPs, or retail/point-of-sale integration. - Experience working with cloud data warehouses (BigQuery, Snowflake, Redshift) and ETL tooling (Airflow, dbt, Kafka). - Excellent written and verbal communication; proven ability to translate technical results to non-technical stakeholders. Preferred qualifications - Experience with incrementality platforms or approaches (e.g., Measured, experimentation platforms, proprietary lift frameworks). - Familiarity with marketing mix modeling (time-series, regularized regression, Bayesian MMM). - Experience with causal ML / uplift modeling and Bayesian inference tools (PyMC3/4, Stan). - Knowledge of identity resolution, privacy-preserving measurement (privacy regulation awareness, cohort-based measurement, differential privacy concepts). - Experience deploying models to production (Docker, CI/CD, MLOps patterns) and instrumenting model monitoring. - Background working with brand/performance media, cross-channel measurement, and agency/client workflows. KPIs / success metrics - Number of incrementality tests designed, executed, and turned into action. - Accuracy and explainability of models (e.g., holdout prediction error, calibration). - Time-to-insight (from data ingestion to stakeholder-ready report). - Business impact (measured media savings or ROI improvements attributable to recommendations). - Adoption rate of measurement outputs by media planners/clients. Nice-to-have (company fit & soft skills) - Proven consultative experience with clients or internal stakeholders. - Comfort operating in ambiguous environments and balancing speed vs. statistical rigor. - Ability to mentor junior data scientists and evangelize measurement best practices across teams.

• Supervise and support participants throughout the practice-oriented modules during the entire duration of the training • Deliver instruction in the online classroom for the “Certified Data Scientist” program • Provide assistance with technical questions and offer guidance to ensure learning success • Give feedback on learning progress, formulate constructive feedback, and actively foster the learning process • Contribute your own ideas and subject-matter expertise to continuously optimize the training concept