Mahdi Kooshkbaghi
Staff Machine Learning Engineer
The Estée Lauder Companies
Biography
As a seasoned machine learning engineer and computational scientist with 10+ years of experience spanning production user-facing AI applications and scientific computing, I currently specialize in building and deploying GenAI-powered products at scale. I have extensive prior research experience in scientific computing, CFD simulation, nonlinear dynamics, and data-driven discovery of governing equations, with a strong track record of developing ML simulation surrogates for complex physical systems. My technical expertise spans deep learning, Bayesian modeling, and scalable MLOps.
You can find my current CV here.
Interests
- Applied Mathematics
- Machine Learning
- Computational Biology
- Combustion
- Fluid Mechanics
Education
Computational Postdoc in Simons Center for Quantitative Biology, 2020
Cold Spring Harbor Laboratory
Postdoc in App. and Comput. Math, 2017
Princeton University
PhD in Mechanical Engineering, 2015
ETH Zürich
MSc in Mechanical Engineering, 2011
Tehran Polytechnic
BSc in Aerospace Engineering, 2010
Tehran Polytechnic
BSc in Mechanical Engineering, 2008
Tehran Polytechnic
Experience
Staff Machine Learning Engineer
- Core developer of the Jo Malone AI Scent Advisor, a user-facing GenAI recommendation system deployed to production. Built question-based recommendation logic and scenario planning using LLM integration on GCP infrastructure.
- Architected and deployed production ML pipelines using Kubeflow on GCP for causal inference modeling and Bayesian estimation with JAX/NumPyro frameworks, including GPU-accelerated MCMC workloads.
- Built containerized ML applications (Docker, Cloud Run, FastAPI) with modular software design, production monitoring, and CI/CD automation via GitHub Actions.
- Mentored interns on interpretable ML projects, leading to a 100% conversion rate to full-time roles, with two now reporting directly to me.
Computational Postdoc
- Developed TensorFlow-based software to analyze Massively Parallel Reporter Assay (MPRA) data.
- Implemented Bayesian inference models for drug interaction analysis using CPU/GPU computing.
- Developed computational frameworks for processing and analyzing raw biological datasets.
Research Fellow in Applied Mathematics
- Led physics-informed machine learning research in collaboration with Johns Hopkins and Brown Universities.
- Developed manifold learning algorithms to analyze time-dependent datasets and extract governing equations.
- Mentored graduate students in developing manifold learning and machine learning software.
Publications
(2024).
Characterization of partial wetting by CMAS droplets using multiphase many-body dissipative particle dynamics and data-driven discovery based on PINNs.
Journal of Fluid Mechanics , Volume 985, A7.
(2024).
Specificity, synergy, and mechanisms of splice-modifying drugs.
Nature Communications 15, 1880 (2024).
(2022).
Machine-learning-based data-driven discovery of nonlinear phase-field dynamics.
Physical Review E 106 (065303) (2022).
(2022).
On the parameter combinations that matter and on those that do not: data-driven studies of parameter (non)identifiability.
PNAS Nexus (2022).
(2022).
Structural and mechanistic basis of σ-dependent transcriptional pausing.
Proceedings of the National Academy of Sciences 119 (23) (2022).
(2022).
MAVE-NN: learning genotype-phenotype maps from multiplex assays of variant effect.
Genome Biology 23, 98 (2022).
(2021).
Novel tool to quantify with single-cell resolution the number of incoming AAV genomes co-expressed in the mouse nervous system.
Gene Therapy (2021): 1-6.
(2021).
Numerical simulation of atomic layer deposition for thin deposit formation in a mesoporous substrate.
AIChE Journal (2021): e17305.
(2021).
Periodicity Scoring of Time Series Encodes Dynamical Behavior of the Tumor Suppressor p53.
IFAC-PapersOnLine 54, no. 9 (2021): 488-495.
(2020).
Emergent spaces for coupled oscillators.
Frontiers in Computational Neuroscience 14 (2020): 36.
(2020).
Manifold learning for organizing unstructured sets of process observations.
Chaos: An Interdisciplinary Journal of Nonlinear Science 30, no. 4 (2020): 043108.
(2020).
Coarse-scale PDEs from fine-scale observations via machine learning.
Chaos: An Interdisciplinary Journal of Nonlinear Science 30, no. 1 (2020): 013141.
(2019).
Generalized entropy production analysis for mechanism reduction.
Combustion Theory and Modelling 23, no. 2 (2019): 197-209.
(2019).
Manifold learning for parameter reduction.
Journal of computational physics 392 (2019): 419-431.
(2016).
Spectral quasi-equilibrium manifold for chemical kinetics.
The Journal of Physical Chemistry A 120, no. 20 (2016): 3406-3413.
(2015).
n-Heptane/air combustion in perfectly stirred reactors: Dynamics, bifurcations and dominant reactions at critical conditions.
Combustion and Flame 162, no. 9 (2015): 3166-3179.
(2015).
On the model reduction for chemical and physical kinetics.
ETH Zurich.
(2014).
The global relaxation redistribution method for reduction of combustion kinetics.
The Journal of chemical physics 141, no. 4 (2014): 044102.
(2014).
Entropy production analysis for mechanism reduction.
Combustion and flame 161, no. 6 (2014): 1507-1515.
(2014).
Non-perturbative hydrodynamic limits: A case study.
Physica A: Statistical Mechanics and its Applications 403 (2014): 189-194.
(2013).
A collocated grid, projection method for time‐accurate calculation of low‐Mach number variable density flows in general curvilinear coordinates.
International Journal for Numerical Methods in Fluids 72, no. 3 (2013): 301-319.
(2010).
A similarity solution in order to solve the governing equations of laminar separated fluids with a flat plate.
Communications in Nonlinear Science and Numerical Simulation 15, no. 12 (2010): 3965-3973.