Projects
Five things that span the range I work in — a reproducible research pipeline, an education platform in daily use, two subscription products, and a symbolic-mathematics toolkit. Everything else lives on GitHub.
01 · Research
BraTS U-Net Research Pipeline
A reproducible, leakage-safe evaluation pipeline for U-Net-family models on multimodal brain-tumor MRI, built for the TÜBİTAK 2209-A project.
- Problem
- Segmentation results in medical imaging are hard to trust when train/test splits leak between patients, evaluation protocols shift between papers, or preprocessing choices go undocumented.
- Approach
- A fixed patient-level split, an audited dataset, and a written analysis plan frozen before the held-out test set was touched — then one guarded internal evaluation against it.
- Engineering
- Standard U-Net, BU-Net and U-Net+RES compared under a bounded, matched 2D protocol; multi-seed confirmation for stability; measured compute per model; every figure and table generated from tracked artifacts; the whole thing verified from a clean clone.
- Result
- Seed-aggregated mean regional Dice on the 74-patient held-out subset: 0.736 standard U-Net, 0.752 BU-Net, 0.756 residual U-Net. Both variants beat the baseline under Holm-corrected paired bootstrap tests; the gap between the two finalists was small and borderline. Residual U-Net got there with fewer parameters, fewer per-slice MACs and lower peak VRAM. Internal single-dataset evidence — no external validation, no clinical claim.
TensorFlowKerasPythonpytest
02 · Production
Online Dershanem
A role-based education and business-operations platform connecting sales, learning delivery, assessment and reporting in one product.
- Problem
- Running exam-prep education — lessons, mock exams, CRM and operational reporting — across disconnected tools creates friction between teaching, sales and operations.
- Approach
- One platform, role-based access: lesson delivery and progress tracking, a mock-exam club, and CRM/ops reporting behind a single account system.
- Engineering
- CI and Lighthouse performance checks on every change via GitHub Actions; Playwright end-to-end tests across browsers; the app ships as a container image to GHCR.
- Result
- Live in production at onlinedershanem.com.
Next.jsTypeScriptPrismaPostgreSQLPlaywrightDocker
03 · Production
OneRead
A subscription product delivering one human-reviewed article briefing every weekday, with verified sign-up and billing.
- Problem
- Daily reading recommendations are either firehoses or unreviewed algorithmic slop; the useful version is one thing, checked by a person.
- Approach
- A single weekday briefing behind email-verified sign-up and subscription billing, with a deliberately small product surface.
- Engineering
- Email-code verification and Polar billing, plus an explicit no-payment test path so the flow is testable end to end.
- Result
- Live production product.
Next.jsTypeScriptPrismaPolar
04 · Production
OneGames
A daily-games sibling to OneRead — OneSudoku, OneDna and OneWord under one shared product shell.
- Problem
- Extend the same product family into daily casual games without fragmenting the brand or rebuilding the shell from scratch.
- Approach
- Reused OneRead's design system — typography, layout, lockup, footer — so both products read as one company, while shipping three independent daily puzzles.
- Engineering
- Verification and subscription billing shared across games; consistent page padding and lockup placement verified by tests.
- Result
- Live production product, sibling to OneRead.
Next.jsTypeScriptPrismaPolar
05 · Research tooling
Quantum Groups
Symbolic Python verification of quantum group structures — representations, Hopf axioms and Yang–Baxter relations.
- Problem
- Verify, by computation rather than by hand, the algebraic claims behind a piece of mathematical research on quantum group structures.
- Approach
- A tightly scoped symbolic toolkit for Uq(sl₂) and GLq(2|1) — not a general theorem prover, but a way to pin the paper's finite-dimensional representations and relations to reproducible tests.
- Engineering
- Explicit E, F, K, K⁻¹ matrices for finite-dimensional representations; representation-level zero-residual checks of the defining relations; Hopf-structure axiom checks (comultiplication, counit, antipode); a 9×9 R-matrix and graded Yang–Baxter verification for GLq(2|1).
- Result
- A reproducible, test-verified companion codebase to the underlying mathematical research.
PythonSymPypytest
← Home
Istanbul · GMT+3