🔭 I’m currently building artificial intelligence and machine learning workflows for materials discovery, linking deep learning models with Density Functional Theory (DFT), Many Body Perturbation Theory based GW calculations, machine learned interatomic potentials, and molecular dynamics simulations. A lot of my current work explores how Bayesian optimization and active learning can reduce the cost of high fidelity electronic structure calculations and accelerate screening of candidate materials.
🔭 In parallel, I am developing applied machine learning projects with a focus on forecasting, optimization, and model evaluation that are closer to industry and quantitative use cases. I am particularly interested in using structured neural networks and probabilistic models to support data driven decision making in domains like time series prediction and risk or anomaly analysis.

👯 I’m looking to collaborate on scientific machine learning projects that connect physics based simulations with modern neural architectures. This includes work on graph neural networks for atomistic systems, surrogate models for DFT or GW level properties, and optimization schemes that couple machine learned interatomic potentials with molecular dynamics or Monte Carlo simulations.
👯 I am equally interested in collaborations on practical artificial intelligence and quantitative modeling problems. This covers areas such as robust time series modeling, portfolio style optimization, reinforcement learning for decision support, and building end to end pipelines that turn raw data into reliable predictive signals.

🤝 On the scientific side, I am looking for feedback on scaling deep learning models that sit on top of DFT and GW workflows, including better dataset design, uncertainty quantification, and strategies for combining heterogeneous simulation data. I am also interested in best practices for training and validating machine learned interatomic potentials and other surrogates without losing physical interpretability.
🤝 On the applied side, I would appreciate guidance on industry grade machine learning engineering: production ready data pipelines, monitoring and evaluation frameworks, and deployment patterns that are common in quantitative finance or large scale analytics. Advice on connecting research style prototypes to business or product metrics would be especially valuable.

🌱 For materials and physics, I am deepening my understanding of graph neural networks for atomic structures, active learning for expensive simulations, and uncertainty aware models that can sit on top of Density Functional Theory and Many Body Perturbation Theory based calculations. I am also exploring improved optimization strategies for tuning parameters such as Hubbard U and exchange mixing in hybrid functionals.
🌱 For quant and applied artificial intelligence, I am learning advanced time series techniques, probabilistic forecasting, and reinforcement learning concepts that relate to sequential decision making. I am also studying practical numerical optimization methods and evaluation protocols that are widely used in quantitative research groups and machine learning teams.

💬 On the scientific computing side, you can ask me about Density Functional Theory workflows, GW quasiparticle corrections, pseudopotential generation, machine learned interatomic potentials, Bayesian optimization for parameter tuning, and how to build automated pipelines that connect electronic structure codes with machine learning models.
💬 On the applied machine learning side, you can ask me about model selection and evaluation, optimization of loss functions, experiment design for ML, and how ideas from physics and simulation can inspire better inductive biases in neural networks for real world data.

⚡ Fun fact: My path runs from electronics and communication engineering to theoretical physics, Density Functional Theory, and Many Body Perturbation Theory, and now into deep learning driven materials discovery. I enjoy translating complex physical insight into models that can actually run on a cluster and produce useful predictions.
⚡ At the same time, I like to treat new quantitative or applied ML problems as an experimental playground, where I can reuse ideas from simulations such as careful uncertainty handling, reproducible workflows, and principled optimization. Bridging these two worlds is slowly becoming my favorite long term project.