Sfoglia categorie
Esplora
Fiverr Pro
Italiano
$
USD
Sunia T
@s_tanweer
PhD Data Scientist for Forecasting, Machine Learning and Statistics
Stati Uniti
Urdu, Inglese
Chi sono
Researcher @ University of Michigan with PhD in Mechanical Engg and Computational Mathematics. I help organizations solve complex engineering and data problems using machine learning, statistics, predictive modeling, and finite element analysis. Expertise includes time-series forecasting, classification, uncertainty quantification, signal processing, feature engineering, machine learning, deep learning, and stochastic systems. I build reliable, validated models that turn complex data into actionable insights.
Python (PyTorch, TensorFlow, scikit-learn, Pandas, NumPy, XGBoost, LightGBM, MLflow)
Competenze
Sunia T
offline •
Consulta i miei servizi
Deep learning
I will build a custom machine learning or deep learning model in python
Machine learning
I will build accurate time series forecasting models for your business
Portfolio
Esperienza lavorativa
Research Scientist
University of Michigan • Full time
Jun 2026 - Present • 2 mos
• Develop statistical and machine learning algorithms for biomedical and scientific datasets within the Michigan Center for Applied and Interdisciplinary Mathematics (MCAIM). • Design mathematically grounded methods for analyzing noisy, high‑dimensional data and collaborate with interdisciplinary researchers in healthcare and AI.
Deep Learning Researcher
Gtech • Part time
Jun 2025 - May 2026 • 11 mos
Building ML and DL classifiers to identify pathogen environments from genomic sequence data using dimensionality reduction, deep neural networks, and off‑the‑shelf LLMs; automating workflows with Github, documenting experiments with MLFlow and deploying models through APIs.
Research Assistant
Michigan State University • Part time
Sep 2022 - May 2025 • 2 yrs 8 mos
• Developed data pipelines and machine/deep learning classifiers like Random Forests and SVM to predict epileptic seizures offline—enhancing accuracy to 99% for single‑channel EEG with topological features, and to 80% for multi‑channel EEG (16% more than traditional features). Used higher‑order causal graph methods for online detection of seizure with 80% accuracy using changepoint detection in complexity and topological metrics. • Automated the detection of change in distribution (data drift) in high‑dimensional stochastic systems by developing novel algorithms (integrating topology, Bayesian spatial modeling, and statistical methods) for robust analysis of noisy time series. • Independently conducted an unsupervised machine learning research project, devising a topology‑based loss function, that achieved up to 20% lower Kullback–Leibler Divergence compared to standard bandwidth selection methods in kernel density estimation with ablation and sensitivity studies. • Mentored 2 first‑year PhD students to get 100% success in qualifying exams; helped write 2 research proposals winning over $600K+ in grant money.
