Name: Eduardo G Gusmao
Role: Senior Machine Learning Researcher | Applied Scientist
Contact: eduardo@gusmaolab.org | English, Portuguese, German, Spanish

Summary: Senior Research Engineer with 8+ years of experience building scalable ML systems for genomics, multi-omics, and biomedical data, bridging research prototypes and production-grade pipelines across large clinical datasets.

Core_Expertise:
  - "Machine Learning"
  - "Deep Learning"
  - "Statistical Modeling"
  - "Bioinformatics"

Career_Profile:
  - "Method Development"
  - "Translational Modeling"
  - "Production-Aware Research"

Professional_Engagements:

  "Technical University of Munich":
    - Title: Senior Machine Learning Researcher
    - Date: Apr 2024 - Current
    - Type: On-Site
    - Role_&_Scope: 
      - Lead a multidisciplinary phenotype + EHR AI initiative within a national-scale precision-medicine program spanning Germany and Brazil, defining technical direction, delivery cadence, and quality standards.
      - Coordinate a distributed team of clinicians, biologists, and ML engineers to design, implement, and operate AI-assisted ETL, harmonization, and data-quality pipelines for heterogeneous real-world clinical data.
      - Architect end-to-end workflows across Python, R, SQL/NoSQL, HPC clusters (SLURM/OpenPBS), and containerized deployments (Docker/Kubernetes), with reproducible environments, automated testing, and CI/CD release governance.
      - Act as the primary technical interface across medical, biological, and computational stakeholders, translating clinical and research requirements into production-grade specifications and maintainable system components.
    - Key_Outcomes:
      - Delivered the core platform approximately 8 months ahead of schedule, unblocking early model training, stakeholder validation, and phased rollout to downstream teams.
      - Achieved multi-order-of-magnitude throughput improvements over prior lab-bound workflows through schema standardization, automated validation, and optimized HPC execution.
      - Enabled large-scale genotype-phenotype analyses across tens of thousands of individuals by providing harmonized feature stores, provenance tracking, and analysis-ready datasets for precision-medicine studies.

  "University of Turku":
    - Title: Senior Machine Learning Engineer
    - Date: Mar 2022 - Apr 2024
    - Type: On-Site
    - Role_&_Scope: 
      - Designed and delivered an end-to-end spatial single-cell RNA-seq analysis platform, spanning laboratory inputs, scalable computation, and user-facing analytical outputs.
      - Led a senior team of bioinformaticians and computer scientists to develop and validate deep-learning-based evaluation frameworks for complex biological signals.
      - Orchestrated heterogeneous production stacks across Python, Docker, JavaScript, and GPU-enabled HPC clusters into a unified, reproducible workflow.
    - Key_Outcomes:
      - Delivered a production-ready, end-to-end pipeline in under 6 months, achieving strong internal adoption across collaborating research groups.
      - Enabled methodological advances supporting T/NK-cell cancer immunotherapy research through robust spatial and single-cell modeling frameworks.
      - Awarded Venia Docendi (Docentship) following extensive teaching, independent evaluation, and demonstrated academic leadership.

  "Göttingen University Medical School":
    - Title: Senior Bioinformatics & Machine Learning Engineer
    - Date: Jan 2020 - Mar 2022
    - Type: On-Site
    - Role_&_Scope: 
      - Unified fragmented ETL and data-quality processes into a single automated platform, establishing reliable ingestion, validation, and auditability across heterogeneous biomedical datasets.
      - Eliminated redundant and unsafe data flows through systematic automation, schema validation, and defensive pipeline design.
      - Built and co-maintained internal data and analytics frameworks using Python and multi-model database systems to support scalable research and operational workloads.
    - Key_Outcomes:
      - Delivered ~60% productivity gains across data-processing workflows by reducing manual intervention and pipeline fragmentation.
      - Reduced wet-lab material consumption and operational costs through earlier error detection, reproducible preprocessing, and tighter data-experiment feedback loops.
      - Led technical and outreach initiatives during the COVID period, supporting institutional response and cross-team coordination.

  "Dana-Farber Cancer Institute":
    - Title: Senior Bioinformatics Scientist
    - Date: Jan 2018 - Jan 2020
    - Type: On-Site
    - Role_&_Scope: 
      - Developed integrated computational and experimental pipelines supporting CRISPR-based target discovery for cancer immunotherapy.
      - Co-authored and maintained strategic bioinformatics tools adopted in hospital and translational research settings.
      - Collaborated across major academic and clinical institutions, including Harvard T.H. Chan, Massachusetts General Hospital, and the Broad Institute, aligning computational analyses with clinical and experimental priorities.
    - Key_Outcomes:
      - Contributed to target-discovery efforts supporting LAG-3-based immunotherapy programs, which advanced into Phase II/III clinical trials with reported objective response rates (ORR) >50% and complete responses (CR) in ~16% of patients.
      - Work contributed to intellectual property later licensed as Relatlimab-rmbw, acquired by Bristol Myers Squibb for approximately USD 3.5M.
      - Contributed to the development of Relatlimab (commonly combined with Nivolumab), now approved and marketed as Opdualag, supporting improved outcomes for patients with advanced cancers.

  "Harvard T.H. Chan School of Public Health":
    - Title: Machine Learning Researcher
    - Date: Jun 2015 - Jan 2018
    - Type: On-Site
    - Role_&_Scope: 
      - Applied deep-learning methods to tumor microenvironment modeling and immunotherapy response prediction, integrating experimental and computational signals.
      - Introduced agile development practices (Scrum-style iteration, backlog prioritization, reproducible milestones) into computational research workflows.
      - Developed and operated Python-based machine-learning pipelines on HPC infrastructure, supporting large-scale model training and evaluation.
    - Key_Outcomes:
      - Authored a novel deep-learning-based immunotherapy biomarker that increased effective CRISPR/Cas9 screening throughput by ~40% through improved off-target filtering.
      - Reduced wet-lab experimentation costs by over 60% by computationally prioritizing high-confidence targets ahead of experimental validation.
      - Enabled the study of multiple novel immunotherapy targets (including LAG-3, B7-H3/BCMA, and CD3), establishing a computational foundation for subsequent NK-cell-focused immunotherapy research.

  "RWTH Aachen University":
    - Title: Machine Learning Researcher
    - Date: Sep 2011 - Jun 2015
    - Type: On-Site
    - Role_&_Scope: 
      - Conducted machine-learning and deep-learning research using Python-based pipelines and HPC infrastructure to address large-scale biomedical and regulatory genomics problems.
      - Designed, trained, optimized, and maintained ML/DL models using reproducible, performance-aware best practices across the full model lifecycle.
      - Authored and maintained a comprehensive regulatory genomics and epigenomics analysis framework supporting large-scale integrative studies.
    - Key_Outcomes:
      - Developed a bias-reduction and training-optimization framework that reduced model training time by ~20% without performance degradation.
      - Co-authored evaluation tooling that improved model performance by ~5% (AUPRC), supported by robust statistical comparison using Friedman-Nemenyi post-hoc testing.
      - Introduced production-grade experiment monitoring (Prometheus, Grafana), increasing team visibility and improving overall productivity by ~30%.
      - Delivered regular deep-learning seminars with sustained participation, consistently positive feedback, and zero course drop-out.

  "Informatics Center (CIn/UFPE)":
    - Title: Machine Learning Engineer
    - Date: Oct 2007 - Sep 2011
    - Type: On-Site
    - Role_&_Scope: 
      - Built and operated machine-learning pipelines and ETL preprocessing workflows on a self-hosted HPC cluster, supporting data-intensive research workloads.
      - Designed and implemented a robust ETL backbone (pre-cloud / pre-AWS) that established durable data-ingestion and preprocessing standards adopted across research groups.
      - Introduced cluster-wide monitoring and observability using Prometheus and Grafana, and delivered hands-on training sessions adopted by multiple teams.
    - Key_Outcomes:
      - Increased lab productivity by ~50% and processing throughput by ~30% by introducing Scrum-based planning and execution into technical workflows.
      - Co-authored data-handling protocols that reduced data leakage and redundancy by ~35%.
      - Reduced unsafe and cyclic ETL patterns by ~45% through systematic automation, validation, and pipeline restructuring.

Education:
  - PhD2: 2015 - 2018 | Harvard University Medical School (Boston, USA) | Biomedical Informatics | Summa Cum Laude
  - PhD1: 2011 - 2015 | RWTH Aachen University (Aachen, Germany) | Life Sciences | Summa Cum Laude
  - MSc: 2010 - 2011 | Federal University of Pernambuco (Recife, Brazil) | Machine & Deep Learning | Grade 100%
  - BSc: 2006 - 2010 | Federal University of Pernambuco (Recife, Brazil) | Computer Science | Grade 89.90%

Development_Environment:
  - Hardware: ["AMD", "ARM", "NVIDIA", "Intel"]
  - OS: ["MAC OS X", "Ubuntu", "Debian", "Fedora", "Windows"]
  - Infrastructure:
    - "Bare-Metal Servers"
    - "VMs"
    - Cloud_Computing: ["AWS", "GCP", "Azure"]
    - HPC_Paradigm: ["Slurm", "OpenPBS", "MPI"]
    - Infra_as_Code: ["Terraform", "CloudFormation", "Pulumi"]

  - Languages:
    - Multi_Paradigm: ["Python", "C/C++", "R", "Bash/Shell", "Julia", "Go", "Rust", "Kotlin"]
    - Web_OO: ["TypeScript", "Java", "Javascript", "C#", "Ruby", "PHP"]
    - Markup: ["YAML", "Quarto", "LaTeX", "HTML/CSS/Markdown"]
    - Declarative: ["SQL", "HCL"]
  - Runtimes: ["CPython", "JVM", "Node.js"]

  - ML_Stack:
    - Frameworks: ["PyTorch", "JAX", "TensorFlow", "Keras", "Hugging Face", "NLTK", "Scikit-Learn"]
    - Engines: ["Spark", "Ray", "TensorRT"]
    - Models: ["Generative Models", "Variational Inference", "Graph Neural Networks", "Attention Hypergraph"]

  - Data_Stack:
    - Databases:
      - SQL: ["PostgreSQL", "MySQL"]
      - NoSQL: ["MongoDB", "Arangodb", "Databricks"]
      - Vector: ["Pinecone", "FAISS"]
    - APIs: ["REST", "GraphQL", "gRPC"]
    - Data_Software: ["Power BI", "Microsoft Suite", "HDF5/Parquet/Zarr"]
    - Data_Tools:
      - Basic: ["Pandas", "NumPy", "Scipy", "Bioconductor", "PySAM"]
      - Big_Data: ["Dask", "Polars"]
      - Specialized: ["PyCaret", "OpenCV"]

  - Web_Stack:
    - Frameworks: ["Django", "React", "Next.js", "Express.js"]
    - Dashboards: ["Dash", "Streamlit", "Gradio"]

  - Systems:
    - Version_Control: ["Git", "Github"]
    - Packaging: ["pip", "(micro)mamba", "(mini)conda", "poetry", "npm"]
    - Containers: ["Docker", "Singularity", "Podman"]
    - Orchestration: ["Kubernetes", "Helm"]
    - CI_CD: ["GitHub Actions", "Jenkins", "GitLab CI"]
    - Observability: ["Prometheus", "Grafana"]