Behind the systems
Requirements, data models, tradeoffs, and reproducible checks.
CASE STUDY 01 · FLORAL STATION DELI BUSINESS PROJECT
Order operations for Floral Station Deli
I implemented the order operations project for Floral Station Deli with a POS connection, server-saved orders, fulfillment workflows, search, and activity history. The version available here is an isolated public demonstration of the workflow. It uses fictional orders and does not expose the business system, customer records, or POS credentials.
Requirements
- Create an order from POS, Web, or Delivery with an exact monetary value.
- Move orders through received, preparing, ready, and fulfilled states; allow cancellation before fulfillment.
- Retain history and restore the queue after a reload.
- Keep each anonymous visitor’s demo records separate.
Database design
HttpOnly browser cookieRandom UUID · 30-day access1 → 0..many
id · primary keyworkspace · indexed scopereference · channel · centsstatus · created_at · updated_athistory · JSON event list1 → 1..many
Embedded in order.historyat · timestampstatus · workflow stateDecisions and tradeoffs
- Integer cents: avoid floating-point arithmetic when storing money.
- Server validation: reject invalid amounts and skipped or backward workflow transitions.
- Optimistic concurrency: updates require the previously read status, preventing a stale tab from overwriting newer progress.
- Embedded history: update the status and its bounded event history atomically in one row. A separate events table would be better for large-scale audit reporting.
- Anonymous demo scope: no recruiter login is required, but clearing the cookie loses access. This is not account-based authentication or a production customer-data system.
- Capacity: a workspace is limited to 200 orders. This limit belongs to the public demonstration; its anonymous access model does not describe the private business deployment.
Testing evidence
The source includes executable validation and SQLite integration checks covering invalid values, lifecycle rules, workspace isolation, durable reloads, and stale updates. These automated source-level checks demonstrate behavior in the public implementation. The business POS connection is described from my implementation experience; these test results do not independently verify that private connection.
Read the recorded test run →CASE STUDY 02 · RELATIONAL DESIGN EXERCISE
University enrollment model
A student can enroll in multiple course sections, and a section can contain multiple students. An enrollment entity resolves that many-to-many relationship. This is an illustrative portfolio model, not a reproduction of the submitted IT 117 assignment.
student_id · PKemail · UNIQUE1 → 0..many
enrollment_id · PKstudent_id · FKsection_id · FKUNIQUE(student_id, section_id)grade · nullable0..many ← 1
section_id · PKcourse_id · FKterm0..many ← 1
course_id · PKtitle · creditsWhy this structure?
Course describes the catalog entry; section describes a specific offering. Enrollment stores the grade because the grade belongs to one student’s attempt in one section, not to the student or course alone. A unique student/section pair prevents duplicate enrollment without preventing a student from retaking a course in a different section.
Validation
The test fixture enables foreign keys, rejects orphan and duplicate enrollments, allows a course retake in a different section, and verifies an enrollment join. The fixture deliberately omits instructors and departments to keep the scope clear.
Download the SQL model →FLORAL STATION DELI · TROUBLESHOOTING
Reconciling ordering and POS product data
Product names, prices, and item configurations were not always consistent between the online ordering platform and POS system. I compared the product records across platforms, identified differences, corrected inconsistent records, and improved the validation process using Excel-based checks.
- Investigate: compare the affected product listings and how their data is organized; look for missing items, inconsistent prices, duplicate entries, and configuration differences.
- Correct: isolate the inconsistent records and make targeted changes.
- Verify: compare the updated information across systems and use spreadsheet checks to confirm alignment.
This experience reinforced a systematic debugging process: reproduce the issue, isolate the cause, make a focused correction, and verify the result.
Python and Pandas automation
I used Python scripts with Pandas to process Excel inventory and vendor data, compare vendor pricing, identify inventory discrepancies, and flag products that needed restocking. I checked results against business records and smaller datasets before relying on them operationally.
Illustrative technical practice
The downloadable Python mapping example and support runbook use fictional records to demonstrate an investigation approach. They are separate from the product-data issue described above and are not exported business code or historical incident records.
CASE STUDY 03 · PERSONAL PROJECT
Security incident triage
An impossible-travel scenario combines two sign-ins seven minutes apart and a denied MFA challenge. The workflow connects identity evidence to containment, audit preservation, recovery verification, and preventive actions.
- Investigate: confirm timestamps, locations, device context, and user activity before deciding how to respond.
- Contain: recommend session revocation, credential reset, and preservation of the audit trail.
- Validate: in a real response, verify session revocation and legitimate user access. The public scenario only changes a simulated state and preset risk score.
- Communicate: summarize evidence, impact, actions, ownership, and follow-up checks.
DEVELOPMENT AND VALIDATION
How I work with code and AI
Git and GitHub
I use version control for my portfolio and personal development projects: tracking changes, staging files, creating commits, pushing to remote repositories, and reviewing changes before deployment. The change history helps recover an earlier working version when needed.
Unix through macOS Terminal
My command-line experience includes navigating directories, managing project files, checking permissions, running development commands, working with Git, and troubleshooting project configurations before deployment. This experience is primarily in development environments rather than production Linux server administration.
ChatGPT and independent validation
I use ChatGPT to support coding, design decisions, troubleshooting, and automation. For Floral Station Deli, I have explored AI-assisted inventory tracking, spreadsheet automation, and product-data validation.
One validation risk is incorrect product matching or an assumption that two spreadsheet columns contain equivalent information. I inspect the matching logic and column meanings, compare outputs against business records, test smaller datasets, and correct discrepancies before operational use.
The public SQL model and recorded checks demonstrate data relationships, integrity constraints, order lifecycle validation, and workspace isolation. These runnable examples demonstrate project behavior; they do not independently verify private business operations.
Business result versus demo result
The deli’s 32% sales increase and 18% waste reduction are results reported in my résumé. The order app’s test results demonstrate software behavior; they do not measure or verify those business outcomes.