Sequential Variational Inference Experiments
As part of a DATA5925 group research project, I contributed Python experiments for sequential variational inference. The work starts from an exact Bayesian linear-regression baseline and compares batch orders, mean-field and full-covariance approximations, and Gaussian and logistic-regression settings.
My contribution includes reproducibility checks and divergence diagnostics to make the experiment comparisons easier to inspect and repeat.
This example demonstrates statistical programming and reproducible research. It is group coursework, not a publication, a clinical model or an independently delivered customer project. The source repository remains private; this portfolio description shares a non-confidential overview only.
0
19
Order Workflow Reliability and Recovery
I built local order and shipment workflow prototypes with TypeScript/Node.js, SQLite and n8n. The workflow validates incoming events, suppresses duplicate deliveries, limits retries and checks remote state before replaying an uncertain write.
The recorded local verification on 6 October 2026 contains 15 passing scenarios. They cover repeated deliveries, transient failures, dropped responses, persisted state after restart and failed readback. These checks demonstrate local recovery behaviour using synthetic orders and local APIs; they are not a customer production deployment.
A handover can include the workflow, event schema, failure log, recovery instructions and acceptance scenarios. Connecting an actual business API requires checking its contract and testing in the approved environment.
0
23
Excel and CSV Cleaning with Exception Reports
0
0
Python Repository Health Inventory
0
0