John Filhmar Ola.

problemThe company's product suite needed one secure, real-time channel for messaging and push notifications instead of fragmented per-app integrations and an SMS dependency.

my roleSole developer & architect — mobile client, distributed backend microservices, deployment pipeline, and production server operations, all built and run single-handedly.

key decisionsDistributed microservice backend with real-time messaging. Mobile application with push notification delivery. Full CI/CD pipeline with automated security scanning and signed artifacts. Production infrastructure and monitoring owned end to end.

problemOrganisations, both government and private, needed one system to run their entire HR operation — delivered as an isolated per-client instance whose data never mixes with any other client's.

my roleSole developer — built the product end to end at YouCode Technologies.

key decisionsCovers the full employee lifecycle: onboarding, time & attendance, leave, payroll, performance, recruitment, training, records and compliance. Client-specific behaviour — branding, leave policies, single sign-on — is configured by admins in-app, so one product serves many organisations with no code changes. Access is governed by a permission model tied to each organisation's own hierarchy: a person can only ever delegate a subset of the access they hold. Shipped as a self-contained, isolated deployment per client — no shared multi-client data store.

problemEvery local government wanted its own citizen app and admin consoles, but building one per city doesn't scale. The design mandate: every per-city difference is data, never code.

my roleSole developer — resident mobile app, the city and platform web consoles, and the backend, end to end.

key decisionsOne platform serves many cities — a new city is onboarded through configuration and content, not a code branch. Resident app: government services, 311 issue reporting, emergency SOS, news, tourism, transport, evacuation centres and a digital ID — usable offline. A city console for staff (content, maps, dispatch, alerts, submission queues) plus a separate platform console for operating the whole fleet of cities. Live emergency dispatch on a real-time command map, and targeted alerts with per-recipient delivery tracking.

problemOrdinary Kanban tools track where work sits but not who produced how much, force managers to assemble end-of-day updates by hand, and give AI assistants no safe, permissioned way to participate as real team members.

my roleSole developer — built the platform end to end at YouCode Technologies.

key decisionsAI-suggested story-point estimates and due dates on new cards, with human confirmation — only confirmed estimates count toward reporting, so the numbers stay trustworthy. Automated end-of-day reports: a scheduled job compiles each workspace's finished work and delivers it to management over SMS or email at a set local time. A native Model Context Protocol (MCP) interface lets AI assistants read and drive the board as an authenticated team member — with exactly that user's permissions. External intake tokens let other systems post bug reports straight onto a board; a per-workspace role hierarchy governs who can do what.

problemMulti-branch automotive service operations needed real-time monitoring and reliable daily transaction reporting — on infrastructure owned and operated end to end, not rented as a black box.

my roleSole developer, owner & maintainer — application, DevSecOps pipeline, and a self-provisioned VPS I secure, manage and operate myself.

key decisionsSelf-provisioned and self-managed VPS — provisioning, deployment, monitoring and maintenance owned end to end. Multi-branch operations with real-time monitoring and daily transaction reports. Four-workflow CI/CD: continuous integration, security scanning, gated deployment, and automated dependency management. Layered security scanning — static analysis, dependency auditing, secret scanning across full git history, dynamic testing, and container vulnerability scanning. Hardened container runtime: multi-stage builds, production-only dependencies, non-root user, correct signal handling, health checks. Application hardening: security headers, origin allow-listing, tiered rate limiting, scoped CSRF protection, and production error sanitisation. Manual-gated deployments with branch and environment guards; secrets injected at runtime with no credentials written to disk.

problemManual expense logging is high-friction and most budgeting apps expose sensitive financial data. Users needed private tracking that updates itself.

my roleSole developer — Android app, encryption model, and notification-derived transaction tracking.

key decisionsEncrypted user financial data — privacy-by-design for sensitive personal records. Derives income and expense tracking from transaction push notifications, minimising manual input. Daily, monthly and annual budgeting and savings goals with threshold alerts. Roadmap includes optional premium features via in-app purchases.

problemLarge, distributed organisations needed to train and test thousands of learners on modest phones over unreliable networks — where a dropped connection must never lose a learner's work.

my roleSole developer — built the platform end to end at YouCode Technologies.

key decisionsComputer-based testing runs entirely on the device and syncs answers back when a connection returns — nothing is lost to a dropped signal. Offline course player for reading and studying lessons, including video, with no connection. Verifiable micro-credentials anyone can check through a public verification portal. Bulk onboarding of thousands of learners processed in the background, so large imports never stall or time out.

problemA facility's turnstiles and face-recognition cameras each spoke different protocols and needed a reliable, offline-tolerant link to the cloud platform — without opening any inbound firewall holes.

my roleSole developer — designed and built the edge service and its web admin console end to end.

key decisionsOne deployable service per site hosts every hardware family behind a common adapter, replacing separate single-purpose bridges. Accepts scan events from turnstiles and face-recognition cameras, decides entry/exit direction, and forwards them to the cloud durably. Keeps operating while the cloud is unreachable, then replays everything it buffered — nothing is lost. All cloud traffic is outbound-initiated, so sites need no inbound firewall openings; ships its own admin console and live video wall.

problemRural clinics relied on paper records and manual scheduling, limiting record accessibility and efficiency for underserved communities.

my roleDeveloper — patient records and appointment management built for low-resource clinic environments.

key decisionsDigital patient records replacing paper-based systems. Appointment scheduling and management. Built for low-resource clinic environments.

problemFlood-prone communities needed early warning that conveys not just that water crossed a threshold, but how fast it is rising — so residents can judge how quickly conditions are worsening.

my roleSole developer — sensors, embedded firmware, and multi-channel alerting.

key decisionsMulti-channel alerting: audible sirens, warning lights, and GSM SMS dispatch. Tracks both water level and rate of rise per monitored area. Designed for community-level deployment in flood-prone areas.