From questions to evidence
Methods, results, and limits from modeling, data, and software work.
Methods, results, and limits from modeling, data, and software work.
Nine projects across modeling, data analysis, and software systems.
Balancing quality, size, and computation.
Time-frequency features and a two-stream U-Net for classification and denoising.
File size fell 25.49%, distortion fell 70.78%, and 14-class accuracy reached about 95%.
Thermal mismatch and defects in BGA and QFN packaging.
Equivalent material models, finite-element analysis, and sensitivity checks.
Corner displacement 3.82×10⁻² mm, effective CTE 22.86 ppm/°C, and modulus 27.75 GPa.
Balancing cost and emissions under practical constraints.
Savings, clustering, nearest-neighbor construction, and simulated annealing.
A nine-vehicle plan with station, ten-year cost, and daily carbon estimates.
Three-dimensional motion, visibility geometry, and coordinated search.
Geometric screening tests with differential evolution, genetic algorithms, and simulated annealing.
One device achieved 4.9 s of screening; three coordinated UAVs reached 16.1 s.
Million-rating analysis on a three-node Hadoop cluster.
HDFS and Hive for cleaning, joining, and aggregation.
Popular films had steadier ratings; the 3% sample limits broader inference.
A controlled comparison of three classifiers.
Accuracy, confusion matrices, and ROC–AUC on one train/test split.
Random forest performed best with 83.41% accuracy and 0.889 AUC.
An object-oriented command-line information system.
Accounts, guest access, permissions, and lost-and-found record management.
Layered services, file persistence, shared validation, and exception handling.
Public messages, administrator replies, and moderation.
Structured message storage with restricted public reads and submissions.
Supabase Auth, AAL2 TOTP, input checks, a honeypot, and database rate limits.
Inference, classification, and dimension reduction for multivariate data.
Multivariate tests, clustering, discrimination, PCA, and factor analysis.
Compared standardized and raw results to show the effect of scale.