CV

📄 You can also download the PDF version here.

Education

Technical University of Munich — M.Sc. Mathematics in Data Science (Oct 2023 – Present)

• GPA: 1.2, top 2% of cohort
• Scholarships: relAI (2023–25), Deutschlandstipendium 2023/24 (declined)

University of Bath — B.Sc. & M.Math in Mathematics (Oct 2016 – Jul 2020)

• GPA equivalent: 1.0 (converted by uni-assist)
• First Class Honours, top 5%

Publications

• Schuchardt, J.; Dalirrooyfard, M.; Guzelkabaagac, J.; et al.
  Privacy amplification by structured subsampling for deep differentially private time-series forecasting
  OpenReview — ICML 2025 Spotlight (top 2.6%)

Research & Projects

Research Intern, University of Alberta — Multi-Objective Generation in Drug Design (Aug 2025 – Present)

• Implementing Pareto-guided diffusion to steer generation toward desirable properties.
• Integrating Bayesian optimization for candidate nomination in generator–oracle loop.
• Benchmarking vs. scalarization and ParetoDrug, reporting hypervolume metrics.

Learning AI for Dextrous Robots Chair, TUM — Self-Supervised Learning for Robot Grasping (May – Aug 2025)

• Leveraged Point-JEPA for 3D point-cloud features, improving grasp success.
• Improved top-logit RMSE by 26% in low–mid label regimes; aiming for workshop submission.

Data Analytics & Machine Learning Chair, TUM — Differential Privacy for Time-Series Forecasting (Oct 2024 – Jun 2025)

• Derived event- and user-level bounds for DP-SGD with structured subsampling.
• Tightened divergences of Gaussian/Laplace mixtures, unlocking stronger guarantees.

Helmholtz Institute of Computational Biology — Deep Learning for RNA Drug Discovery (Oct 2024 – Apr 2024)

• Predicted RNA–ligand affinities using GNNs and RNA-FM embeddings.
• Applied LoRA fine-tuning, cross-attention pretraining on curated dataset.

Visual Computing & AI Chair, TUM — Zero-Shot 3D Shape Correspondence (Apr – Jul 2024)

• Built GeoAware3D pipeline fusing diffusion + DINO features into 3D descriptors.
• Achieved near-SOTA accuracy on SHREC’19 semantic correspondence with 4× faster inference.

Work Experience

Data Science Graduate — Lloyds Banking Group, London (Sep 2022 – Apr 2023)

• Built enterprise emissions-estimation model with gradient boosting, +15% accuracy.
• Applied synthetic augmentation for sparse SME data; tuned for UK-wide deployment.

Analytics Consultant — Hyper Group, Leeds (Sep 2021 – Jul 2022)

• Designed ETL pipelines from AWS → Snowflake (+25% throughput).
• Created executive dashboards on Cloud SQL, defining KPIs on price elasticity & retention.

Awards & Extracurriculars

• Konrad Zuse School of Excellence in Reliable AI (relAI) — Full Scholarship (2023–25)
• University of Alberta Research Experience (UARE) — Research Stipend (2025, DAAD & BMBF funded)