Explore showcase

ACM Cinema

Scrum Master & Full-Stack · Aug 2025 – Present

Project1 / 13

End-to-end booking lifecycle: discovery/search → seat selection → checkout/payment → order flows.
AuthN/AuthZ with role-based gates for user vs admin capabilities.
Pricing and promotions integrated into booking and payment paths.
Admin workflows for domain management and operational controls.
Reusable frontend domain hooks for movies, search, seats, payments, orders, and users.
UI polish via Framer Motion on promotional and background components.
Pagination and caching to keep data-heavy views responsive.

Designed layered frontend/backend flow to preserve maintainability as feature scope expanded.
Implemented 80+ REST endpoints and 15 JPA entities over PostgreSQL (Neon), with relational modeling and migration iteration.
Integrated BCrypt for one-way password hashing and AES-based reversible encryption in sensitive payment-related flows (with later acknowledgment that this security model needed stricter server authority/tokenization strategy).
Applied facade-like client wrappers and aggregation strategies to standardize integrations across auth/movies/booking/promotions/payments/admin.
Implemented Virtual Proxy/lazy-loading style responses in targeted flows to reduce initial payload overhead.
Used React optimization primitives (`useMemo`, `useCallback`, `useRef`, context composition) to stabilize render behavior in complex pages.
Added Caffeine cache TTL strategy (10 min) and coordinated with query/refetch behavior to manage freshness/performance tradeoffs.
Added pagination and indexing-focused query improvements for frequently accessed records.
Established Jira process, team ticketing discipline, and dev-guide conventions for multi-sprint consistency.

Requirements were strict and sometimes unclear; documentation overhead was large.
Architectural ambition introduced conflicting partial abstractions (multiple wrapper/facade variants with inconsistent semantics).
AI-assisted scaffolding accelerated development but often introduced inconsistent patterns and premature optimization debt.
AuthN/AuthZ was one of the hardest subsystems and created difficult debugging cycles when implementation shortcuts were taken.
Testing maturity lagged behind implementation pace (manual Postman/curl/print-debug heavy), causing regression risk.
Some server authority weaknesses around client-supplied context.
Difficulty with using Flyway for migration with Springboot/Neon.
Duplicated request logic across multiple client wrappers.
Oversized hooks (especially seat orchestration) as complexity hotspots.
Overlapping caching layers (manual cache + query cache + memoization) increasing stale-data invalidation complexity.

This is still my hardest and most ambitious project by far.
The challenge wasn’t just feature count—it was balancing architecture quality, team coordination, docs/process requirements, and all the new patterns I wanted to try.
I learned that trying to implement every “industry-style” optimization at once can backfire if consistency and testing don’t keep up.
AI helped a lot for skeletons and refactor suggestions, but it also made it easier to accidentally introduce inconsistent code and “premature optimization” traps.
Biggest regret: we did not establish stronger automated testing/GitOps early enough.
Still, I’m proud of the grind: tons of self-learning in Spring/React/Postgres, many all-nighters, and real leadership reps through Jira and project structure.
We received strict demo feedback, fixed issues post-demo, and were told it was among the strongest projects in class. We were absolutely cheesing.

ACM Cinema demo

Overview