Dr. Amanda J. Minnich


Senior Applied Machine Learning Researcher


Amanda J. Minnich

Senior Applied ML Researcher,

Azure Trustworthy ML Team, Microsoft

MS + PhD, Computer Science,

University of New Mexico

BA, Integrative Biology,

University of California, Berkeley

  • White LinkedIn Icon
  • White Twitter Icon
  • White Instagram Icon


I am an applied machine learning researcher focused on the adversarial and platform health space. At Microsoft I use adversarial machine learning algorithms to attack Microsoft's ML systems. At Twitter I developed graph clustering algorithms to detect spam and abuse campaigns. Previously I led the Molecular Data-Driven Modeling Team at Lawrence Livermore National Laboratory, where we applied machine learning methods to the drug discovery process. I am also involved with tech outreach efforts, especially for women in tech. Outside of work, I love taking my dogs Nala and Gizmo for hikes, cooking, baking sourdough bread, watching reality TV, traveling with my husband Jayson Grace, and spending time with family and friends. I am a proud New Mexican and currently live in Lakewood, CO, USA.



University of New Mexico – Albuquerque, NM; Class of 2017

PhD; Computer Science; Dissertation  title:  “Spam,  Fraud,  and  Bots:  Improving  the  Integrity  of  Online  Social  Media  Data”

GPA: 4.04/4.0


University of New Mexico – Albuquerque, NM; MS; Computer Science

GPA: 4.06/4.0


University of California, Berkeley; BA; Integrative Biology

GPA: 3.66/4.0

Skills & Languages

Programming Languages, Libraries, and Tools
  • Python (Pandas, sklearn, TensorFlow, Matplotlib, etc.) - 10 YOE

  • SQL (BigQuery, PostgreSQL, MySQL, Presto) - 8 YOE

  • Git - 9 YOE

  • Docker and Kubernetes - 2 YOE

Machine Learning/Data Science Methods
  • Supervised and unsupervised algorithms

  • Classical ML and deep learning

  • Various types of feature selection/pruning

  • Hyperparameter optimization



Summer 2015

Groupon Inc.; Data Science Intern

  • Designed a predictive bid regression model with an expanded feature set for improved SEM ad performance

  • Implemented smart keyword generation for products using NLP analysis of product descriptions.

Summer 2014

Mandiant, a FireEye company; Data Science Research Intern 

  • Wrote malware family random forest classifier that was put into production and is currently part of company's toolkit

  • Modified JavaScript's D3 library's Force Layout to implement a Barnes-Hut approximation of t-SNE

Summer 2013

Center for Cyberdefenders, Sandia National Laboratory; Data Science Research Intern

  • Applied k-means clustering to Frobenius norm inter-year distances for dimension reduction of system call trace-based Markov chain matrices

  • Created random forest classifier model to identify malware


Jan. 2020 – Aug 2021

Twitter Inc., Data Scientist II, Scaled Enforcement Heuristics 

  • I created automated pipelines to detect inauthentic coordinated behavior using unsupervised machine learning methods.

  • My work spanned the full spectrum of research, prototyping, A/B testing, and productionization, as well as firefighting high-priority spam and abuse issues on the platform.


Aug 2021 - Present

Microsoft, Senior Applied Machine Learning Researcher, Azure Trustworthy Machine Learning Team

  • I use state-of-the-art adversarial ML algorithms to compromise Microsoft's ML systems


July 2017 -- Jan 2020

Lawrence Livermore National Lab; Machine Learning Research Scientist, Molecular Data-Driven Modeling Team Lead

  •  I served as the data-driven modeling tech lead for the ATOM Consortium, where we integrated machine learning into the drug discovery process.

  • I was the chief architect for the ATOM Modeling PipeLine, an open source deep learning pipeline, which supports the whole machine learning life cycle: data processing; feature extraction/normalization; model training and evaluation; ad hoc prediction generation; and model/data storage, provenance, and validation. 


Selected Media

Awards & Service to Profession

  • Wogrammer Spotlight (June 2020)

  • Artificial Intelligence Track Co-Chair, Grace Hopper Celebration (2019 and 2020)

  • Co-Organizer, Fifth Computational Approaches for Cancer Workshop at SC (2019)

  • Program Committee Member, KDD19, CSoNet19, ASONAM17, ASONAM18, and ASONAM19

  • President and Co-founder, UNM Women in Computing (2015-2017)

  • Grace Hopper Celebration Scholar, for Outstanding Women in Computer Science (2014)

  • NIH Programs in Biology and Biomedical Sciences Fellow (2013-2015)

  • National Science Foundation Graduate Research Fellow (2012-2017)



  • DEF CON 30, Las Vegas, NV, 2022: "Hands-on Hacking
    of Reinforcement Learning Systems

  • CompBioMed, London, UK, 2019: "Safety, reproducibility, performance: Accelerating cancer drug discovery with ML and HPC technologies."

  • SIAM International Conference on Data Mining, Calgary, Alberta, Canada, 2019: "Taming social bots: Detection, exploration and measurement." With A. Mueen and N. Chavoshi

  • NVIDIA GPU Technology Conference, San Jose, CA, 2019: "Using GPUs to generate reproducible workflows to accelerate drug discovery."

  • HPC User Forum, Santa Fe, NM, 2019: "Safety, reproducibility, performance: Accelerating cancer drug discovery with ML and HPC technologies."

  • Fourth Computational Approaches for Cancer Workshop at SuperComputing, Dallas, TX, 2018: "Safety, reproducibility, performance: Accelerating cancer drug discovery with cloud, ML, and HPC technologies."

  • National Laboratories Information Technology Summit, Nashville, TN, 2018: "Utilizing container technology to streamline data science."

  • 7th Temporal Web Analytics Workshop at WWW, Perth, Australia, 2017: "Temporal patterns in bot activities." On behalf of Nikan Chavoshi.


Journal Papers


Conference Papers