Community work
Workshops lead in the Software & Technology Club; secretary and technical committee member in the Computer Engineering Society.
C++ OOP and Git/GitHub workshopsMehmet Baran Çelebi
From idea to product. End to end.
Demod · output
I build mobile apps, AI products, real-time communication tools, and business systems. I take ideas through design, development, and testing to software people can use.
- Education
- Yaşar UniversityB.Sc. · 2020 — 06.2026
- Internship
- Spartaküs Yazılım A.Ş.ERP · 06.2025 — 06.2026
- Products
- 09 projects3 on Google Play
02CHANNELS · PROJECTS
Selected work
Published mobile apps, live web products, and software systems I have built. Each answers a different problem.
HEARD00 / 09
Channel 01
AI AND BROADCAST SYSTEMS
Kuki
SIGNAL SIGNATURE · ILLUSTRATIVE — SENTENCE-BY-SENTENCE STREAMING VOICE
- Channel
- 01 / 09· Kuki
- Band
- AI products
- Status
- Working broadcast system
A Turkish-speaking AI streamer connecting game telemetry and screen observation with real-time voice, a 3D avatar, and broadcast automation.
- automated tests
- Real-time Turkish conversation
- Game profiles and telemetry
- 3D VRM avatar and lip sync
Channel guide
- On air: Google Play or a live web product
- Built
- In development
- Channel 01KukiAI and broadcast systemsautomated testsBuilt
Kuki — Technical details
Product and problem
It brings game context, live chat, and broadcast events together for real-time voice interaction through a consistent character on Kick.
My contribution
I designed and developed Kuki’s broadcast system, from telemetry integrations and streaming speech to the Unity avatar and operator controls.
System flow
- Game data and screen
- Context and priorities
- Response and streaming voice
- Avatar and broadcast
Testing and scope
- 198 automated tests
Key capabilities
- Real-time Turkish conversation
- Game profiles and telemetry
- 3D VRM avatar and lip sync
- Broadcast automation
- Operator controls
- Viewer memory across streams
Engineering notes
Understanding the game in context
A game-profile system supports adaptation to different games. A custom Fabric mod supplies Minecraft telemetry, while the League of Legends integration uses the Riot Live Client Data API. Computer vision provides screen observation across games. Kuki interprets these inputs and comments on play in context.
Conversation and event handling
OpenAI Realtime generates responses and ElevenLabs delivers sentence-by-sentence streaming speech. Chat messages, gifts, and game events are prioritized within the broadcast flow.
Bringing the character on screen
A VRM character in Unity expresses responses through speech-timed lip sync and emotion-driven gestures and body language. Its reactions connect the voice to the current broadcast context.
Control and broadcast safety
Profanity, spam, and prompt-injection filters handle incoming content. An emergency mute in the operator panel can stop speech when needed. OBS WebSocket controls broadcast scenes and overlays.
Treating the broadcast as a product
Animated chat, polls, one-click startup, and viewer memory across streams form part of the same system. Development is supported by 198 automated tests.
Technology stack
Python · asyncio · aiohttp · Pydantic · OpenAI API · ElevenLabs · Unity / C# · Java / Fabric · OBS WebSocket · HTML / CSS / JavaScript · SQLite · PowerShell
Status
Working broadcast system
Illustrative system concept - Channel 02ConsiliariusAI advisoryfeaturesIn development
Consiliarius — Technical details
Product and problem
It centers the company’s goals, decision patterns, and business context to support more consistent guidance and decision-making.
My contribution
I am designing and developing Consiliarius as a product that adapts to the company it belongs to, develops its advisory persona around that organization’s needs, and supports its decisions.
Key capabilities
- Company-specific advisory identity
- Organizational context adaptation
- Structured decision support
- Evolving advisory experience
Engineering notes
An advisor shaped by the company
The product centers on an advisory persona that develops with the company’s way of working. Organizational context, goals, and decision patterns inform an advisory experience designed to adapt to that company’s needs.
Status
Product in development
Illustrative system concept - Channel 03Mona: AI Dating & MatchmakerAI-enabled mobile productGoogle PlayOn air
Mona: AI Dating & Matchmaker — Technical details
Product and problem
It brings preference discovery, match suggestions, profile and conversation guidance, and safety features into one product flow.
My contribution
I designed, developed, and published the application on Google Play.
Key capabilities
- Preference discovery
- Match suggestions
- Profile and chat guidance
- Safety features
Engineering notes
Guidance within the product flow
AI-assisted preference discovery, profile guidance, and conversation support sit within the same dating experience. The product is available as an Android application on Google Play.
Status
Published on Google Play
Google Play store listing - Channel 04Bubble Shooter: Last StandMobile gameGoogle PlayOn air
Bubble Shooter: Last Stand — Technical details
Product and problem
It extends the classic bubble-shooter loop with power cards and a daily challenge.
My contribution
I built it with Flutter and Dart and released it on Google Play with advertising and in-app purchases.
Key capabilities
- Endless play
- Power cards
- Daily challenge
- Offline play
- Turkish / English
Engineering notes
From gameplay to store delivery
Power cards and a daily challenge extend the gameplay loop. Offline use, Turkish/English support, advertising, and in-app purchases are part of the released mobile product.
Technology stack
Flutter · Dart
Status
Published on Google Play
Google Play store listing - Channel 05Blockivox: Echo PuzzleMobile gameGoogle PlayOn air
Blockivox: Echo Puzzle — Technical details
Product and problem
Placed blocks create echo chains within a puzzle loop with multiple modes.
My contribution
I built it with Flutter and Dart and published it on Google Play with offline play, advertising, and in-app purchases.
Key capabilities
- Echo-chain mechanic
- Multiple modes
- Daily challenge
- Offline play
Engineering notes
Building a product around a mechanic
Echo chains shape the game’s core decision loop. Different modes, a daily challenge, and offline play turn the mechanic into a mobile product, with advertising and in-app purchase integrations.
Technology stack
Flutter · Dart
Status
Published on Google Play
Google Play store listing - Channel 06PandemoniumGame and enginepandemoniumgame.netBuilt
Pandemonium — Technical details
Product and problem
It brings the decision-making and sense of adventure of tabletop role-playing into a turn-based structure designed for a digital game.
My contribution
I developed Pandemonium end to end, from its game design and software to the engine that powers it, shaping both under one product vision.
Key capabilities
- Turn-based game structure
- Tabletop role-playing focus
- Purpose-built game engine
- Integrated game and engine development
Engineering notes
Developing the game and engine together
I approached the game experience and its engine as parts of the same product. The turn-based structure and tabletop role-playing focus guide both game design and software development decisions.
Status
Completed game system
Illustrative system concept - Channel 07M.E.M.O RTCReal-time communicationautomated server testsOn air
M.E.M.O RTC — Technical details
Product and problem
It removes account creation and installation from starting a call, bringing browser-based communication and shared listening into one room.
My contribution
I designed and developed the application, audio-processing pipeline, and room management end to end, and deployed it on Render.
System flow
- Join a room
- Signaling and connection
- Voice, video, and screen
- Communicate together
Testing and scope
- 46 automated server tests
- 1080p screen sharing support
Key capabilities
- Account-free rooms
- Voice and video
- 1080p screen sharing
- RNNoise suppression
- Shared YouTube music
- Mobile PWA
Engineering notes
Connections and rooms
WebRTC handles peer-to-peer voice and video, while Node.js and WebSocket provide signaling and room management. Cloudflare TURN relays media when required. Opus codec settings target low-latency communication.
Resilient signaling
If the signaling server connection drops while the media connection remains available, the existing WebRTC call is preserved and signaling reconnects. This is distinct from losing internet connectivity entirely.
Audio processing on its own thread
WebAssembly-based RNNoise suppression works alongside an environment-adaptive noise gate and an audio limiter. Processing runs on a separate audio-processing thread.
More in the same room
The application supports 1080p screen sharing, video, and front/rear camera switching. Text chat, shared YouTube music listening, and administrator permissions complete the room experience.
From the browser to everyday use
The mobile-friendly PWA can be added to the home screen. It is deployed on Render, with server behavior covered by 46 automated tests.
Technology stack
JavaScript · Node.js · WebRTC · WebSocket · Web Audio API · WebAssembly · RNNoise · HTML / CSS · Render · Cloudflare TURN
Status
Deployed on Render
Illustrative system concept - Channel 08M.E.M.O WMSDesktop warehouse managementautomated testsBuilt
M.E.M.O WMS — Technical details
Product and problem
It gives small and medium-sized manufacturing plants a traceable record of where materials are, which lot they belong to, and their current status.
My contribution
I developed the warehouse workflows, desktop interface, and local data layer, making inventory consistency central through tests and controlled transaction approval.
System flow
- Receive materials
- Warehouse, shelf, and lot
- Production and movements
- Count and reconcile
Testing and scope
- 1,000+ automated tests
- 22 movement types
- 10 built-in reports
Key capabilities
- Warehouse, shelf, and lot tracking
- Immutable inventory ledger
- Keyboard-first entry
- Blind counts and reconciliation
- ERP data exchange
- Automatic backups
Engineering notes
The factory’s material flow
The system supports 22 movement types, including receiving, dispatch, transfers, production issues and receipts, subcontracting dispatch and returns, quality quarantine, and scrap. Warehouse, shelf, lot, and stock status are tracked together.
Fast entry, traceable records
Paper vouchers are entered through an Excel-like, keyboard-first interface. An immutable inventory ledger and atomic approval mechanism protect records; negative balances are not allowed. Corrections use reversal vouchers that preserve the audit trail.
Counts and operational visibility
Saved count snapshots, blind counts, and annotated variance vouchers support stocktaking. Minimum-stock alerts and 10 built-in reports with Excel/PDF exports provide inventory visibility.
Working with an existing ERP
CSV/XLSX imports support product transfers and balance reconciliation. The application runs on the desktop with a local SQLite database, without internet access or a separate server.
Data and version discipline
Built with Rust, Tauri v2, and React/TypeScript, the system uses a layered architecture, versioned database migrations, and automatic backups.
Testing and measurement scope
The project has more than 1,000 automated tests. Performance was measured with a dataset of 10,000 products and 100,000 movements.
Technology stack
Rust · Tauri v2 · React · TypeScript · SQLite
Status
Completed desktop system
Illustrative system concept - Channel 09GelinlikOSWeb SaaS · Team projectgelinlikos.comOn air
GelinlikOS — Technical details
Product and problem
It brings customer records, fittings, atelier tasks, managerial approvals, and delivery into one workflow.
My contribution
We developed GelinlikOS as a team. I contributed to the mobile-first web SaaS product as part of that collaborative development effort.
System flow
- Customer record
- Fittings and atelier
- Manager approval
- Delivery
Key capabilities
- Customer records
- Fitting workflow
- Atelier tasks
- Manager approvals
- Delivery flow
Engineering notes
The atelier’s everyday workflow
Customer records, fittings, and atelier tasks share one process leading to delivery. The mobile-first approach reflects an atelier environment where work is not confined to a desk.
Status
Live web productTeam project
Real product screen
03ROUTING · CAPABILITIES
Tools, in the work itself.
Development capabilities and technologies connected to the products I build.
CapabilityRELATED WORKDEVELOPINGKukiConsiliariusMonaBubble ShooterBlockivoxPandemoniumM.E.M.O RTCM.E.M.O WMSGelinlikOS
Mobile product development
I connect cross-platform interfaces with game loops and app-store delivery.
Flutter · Dart · React Native · Expo · Android
Real-time communication
Browser-based calling, media connections, and audio processing on a separate thread.
Node.js · WebRTC · WebSocket · Web Audio API · WebAssembly · RNNoise
AI products and broadcast systems
I connect AI to products through LLM integration, prompt design, and structured outputs; in Kuki, I bring game context, response generation, streaming speech, and character reactions into a working broadcast flow.
Python / asyncio · OpenAI API · ElevenLabs · Unity / C# · Java / Fabric · OBS WebSocket
Desktop and local data
I model business workflows through traceable movements and consistent local records.
Rust · Tauri v2 · React · TypeScript · SQLite · Database migrations
Services and product delivery
I treat device, service, and deployment integrations as parts of the product.
REST API / JSON · Firebase · SDK integrations · Push notifications · AdMob · Google Play Console / AAB · EAS Build · Render
Software engineering and testing
Debugging, automated testing, and performance measurement are part of my development process.
Git / GitHub · Linux · Automated testing · Performance optimization · OOP · C / C++ / Java · JavaScript / TypeScript · AI-assisted development
Foundations I am developing
I am developing foundational knowledge of cybersecurity, computer vision, model fine-tuning, and RAG through study and prototyping.
Cybersecurity fundamentals · Computer vision fundamentals · Model fine-tuning fundamentals · RAG fundamentals
Frequency list
Every tool, patched to the channel that uses it. Filled: channel · outlined: capability row.
- AdMobServices and product delivery: 05
- AI-assisted developmentSoftware engineering and testing: 06
- aiohttp01
- AndroidMobile product development: 01
- asyncio01AI products and broadcast systems: 03
- Automated testingSoftware engineering and testing: 06
- CSoftware engineering and testing: 06
- C++Software engineering and testing: 06
- Cloudflare TURN07
- Computer vision fundamentalsFoundations I am developing: 07
- CSS0107
- Cybersecurity fundamentalsFoundations I am developing: 07
- Dart0405Mobile product development: 01
- Database migrationsDesktop and local data: 04
- EAS BuildServices and product delivery: 05
- ElevenLabs01AI products and broadcast systems: 03
- ExpoMobile product development: 01
- Fabric01AI products and broadcast systems: 03
- FirebaseServices and product delivery: 05
- Flutter0405Mobile product development: 01
- Git / GitHubSoftware engineering and testing: 06
- Google Play Console / AABServices and product delivery: 05
- HTML0107
- Java01AI products and broadcast systems: 03Software engineering and testing: 06
- JavaScript0107Software engineering and testing: 06
- LinuxSoftware engineering and testing: 06
- Model fine-tuning fundamentalsFoundations I am developing: 07
- Node.js07Real-time communication: 02
- OBS WebSocket01AI products and broadcast systems: 03
- OOPSoftware engineering and testing: 06
- OpenAI API01AI products and broadcast systems: 03
- Performance optimizationSoftware engineering and testing: 06
- PowerShell01
- Push notificationsServices and product delivery: 05
- Pydantic01
- Python01AI products and broadcast systems: 03
- RAG fundamentalsFoundations I am developing: 07
- React08Desktop and local data: 04
- React NativeMobile product development: 01
- Render07Services and product delivery: 05
- REST API / JSONServices and product delivery: 05
- RNNoise07Real-time communication: 02
- Rust08Desktop and local data: 04
- SDK integrationsServices and product delivery: 05
- SQLite0108Desktop and local data: 04
- Tauri v208Desktop and local data: 04
- TypeScript08Desktop and local data: 04Software engineering and testing: 06
- Unity / C#01AI products and broadcast systems: 03
- Web Audio API07Real-time communication: 02
- WebAssembly07Real-time communication: 02
- WebRTC07Real-time communication: 02
- WebSocket07Real-time communication: 02
04 PROCESSING · APPROACH
The work between idea and use.
I think about the whole product, from interface and data flow to the first prototype and delivery.
- RAW SIGNAL
Understand the context
I clarify the user’s task, the actual problem, and the system’s boundaries.
- SHAPING
Build the system
I develop the interface, data flow, and services around a shared product goal.
- VERIFICATION
Verify the behavior
I bring failure scenarios, automated tests, and real-device use into the development process.
- BROADCAST
Deliver and improve
I use deployment, versioning, and feedback to keep the product usable and ready to evolve.
05RECORD · BACKGROUND
Learn. Build. Share.
EDUCATION
—
Yaşar University
B.Sc. Computer Engineering
EXPERIENCE
—
Spartaküs Yazılım A.Ş.
Software Development Intern, ERP Systems
Languages
- TurkishNative
- EnglishAdvanced
- JapaneseBeginner
- RussianBeginner
06BROADCAST · CONTACT
Let’s talk about the next idea.
Email me about a new product, a team opportunity, or a technical idea.
00 / 09 channels heard.