Event operations foundations
Built practical experience in logistics, on-site coordination, team supervision and time-sensitive execution.
diff --git a/.github/workflows/site-preflight.yml b/.github/workflows/site-preflight.yml index a3088d7..38c11f2 100644 --- a/.github/workflows/site-preflight.yml +++ b/.github/workflows/site-preflight.yml @@ -34,6 +34,16 @@ jobs: - name: Prepare reports directory run: mkdir -p qa-results + - name: Check JavaScript syntax + run: | + set -e + for file in ./*.js; do + node --check "$file" + done + + - name: Check portfolio architecture consistency + run: npm run qa:portfolio + - name: Validate HTML continue-on-error: true run: npm run qa:html 2>&1 | tee qa-results/html-validation.txt diff --git a/.pa11yci b/.pa11yci index d733f4b..1c00895 100644 --- a/.pa11yci +++ b/.pa11yci @@ -3,16 +3,8 @@ "standard": "WCAG2AA", "timeout": 30000, "wait": 750, - "viewport": { - "width": 390, - "height": 844 - }, - "chromeLaunchConfig": { - "args": [ - "--no-sandbox", - "--disable-setuid-sandbox" - ] - } + "viewport": { "width": 390, "height": 844 }, + "chromeLaunchConfig": { "args": ["--no-sandbox", "--disable-setuid-sandbox"] } }, "urls": [ "http://127.0.0.1:4173/index.html", @@ -20,6 +12,8 @@ "http://127.0.0.1:4173/games.html", "http://127.0.0.1:4173/sinama-case-study.html", "http://127.0.0.1:4173/merge-rush-case-study.html", + "http://127.0.0.1:4173/labs.html", + "http://127.0.0.1:4173/now.html", "http://127.0.0.1:4173/blog.html", "http://127.0.0.1:4173/single-work.html", "http://127.0.0.1:4173/request.html", diff --git a/PORTFOLIO_ARCHITECTURE.md b/PORTFOLIO_ARCHITECTURE.md new file mode 100644 index 0000000..307143c --- /dev/null +++ b/PORTFOLIO_ARCHITECTURE.md @@ -0,0 +1,167 @@ +# Portfolio Architecture V2 + +## Why this exists + +The portfolio grew from a static personal site into a product with Recruiter Mode, Ajoop, Command Palette, bilingual copy, case studies, games, request intake, accessibility QA and multiple active flagship products. + +The previous structure duplicated current project truth across HTML pages, the monolithic runtime and `flagship-copy.js`. That made drift likely whenever SINAMA, Merge Rush or career positioning changed. + +## Source-of-truth rule + +For new portfolio work, **current product facts, recruiter profiles, build status and evidence summaries live in `portfolio-data.js` first**. + +Do not add a new SINAMA / Merge Rush fact directly to Ajoop or Recruiter Mode without first adding it to the registry. + +### `portfolio-data.js` + +Owns: + +- Profile and contact truth +- Flagship project summaries +- Project status / stack / proof +- Role-specific recruiter profiles +- Build log entries +- Labs catalog +- Sanitized SINAMA Evidence Explorer examples + +### `portfolio-v2.js` + +Owns: + +- Unified EN/TR attribute copy (`data-pv2-en` / `data-pv2-tr`) +- Runtime synchronization of Ajoop with registry facts +- Recruiter Mode V2 rendering +- Role-specific URLs such as `?role=applied-ai` +- Build Log rendering +- Labs rendering +- SINAMA Evidence Explorer rendering +- Browser-side request receipt note +- Command Palette extensions for Labs, Build Log and recruiter deep links + +### `portfolio-v2.css` + +Owns UI for the V2 runtime surfaces. New V2 components should be styled here instead of growing `style.css` unless they are true global primitives. + +## Runtime compatibility layer + +The old monolithic runtime is preserved byte-for-byte as `legacy-script.js`. + +`script.js` is now a small bootloader. + +On pages explicitly migrated to V2, the HTML already declares `portfolio-data.js` before `script.js` and `portfolio-v2.js` after it. The bootloader detects that setup and synchronously inserts only `legacy-script.js`, preserving execution order. + +On older pages that still contain only ``, the bootloader inserts: + +`portfolio-data.js → legacy-script.js → portfolio-v2.js` + +It also adds `portfolio-v2.css` when missing. + +This keeps current Recruiter Mode and Ajoop evidence available on older case studies, certificates, 404 and mini-game pages without forcing a risky all-at-once HTML rewrite. + +Future refactors can extract focused modules from `legacy-script.js`; the compatibility bootloader gives that work a stable migration boundary. + +## Ajoop scope + +Ajoop is deterministic in this release. It maps portfolio questions to curated evidence-backed answers and links. + +V2 improves its grounding by sourcing current SINAMA, Merge Rush, role-fit and build-status facts from the central registry, but **does not claim a live LLM, semantic retrieval or RAG backend**. + +A future semantic assistant should be a separate, explicit product step with source citation, safe fallbacks and clear runtime boundaries. + +## Role deep links + +Supported role ids: + +- `applied-ai` +- `solution-engineering` +- `software` +- `game` + +Example: + +`https://kaanbalci.com/?role=applied-ai` + +A valid role parameter selects that evidence profile and opens Recruiter Mode. + +## Page hierarchy + +### Professional narrative + +1. `index.html` — concise landing page +2. `works.html` — curated professional evidence +3. `sinama-case-study.html` — flagship Applied AI proof +4. `merge-rush-case-study.html` — flagship game / interactive product proof +5. `blog.html` — professional timeline +6. `about.html` — profile and capability map +7. `request.html` — project inquiry + +### Supporting surfaces + +- `now.html` — living build log +- `labs.html` — technical experiments that should not dilute the main career story +- `games.html` — actual game / playable work +- `project-detail.html` — dynamic archive detail route; intentionally `noindex` +- `single-work.html` — training / certificates + +## Evidence policy + +Prefer evidence over quantity metrics. + +Good proof: + +- Implemented product behavior +- Test suites / regression logic +- Release-readiness rules +- Real customer workflows +- Responsive QA evidence +- Architecture boundaries +- Live product links + +Weak proof should not lead the homepage: + +- Tutorial-scale repositories +- Calculator / weather practice apps +- Unbuilt concepts +- Raw certificate counts + +These can remain in GitHub, Labs or archive surfaces. + +## SINAMA Evidence Explorer + +The explorer uses simplified, sanitized static examples. It is not a live evaluator and must never be presented as one. + +Its purpose is to make the product model understandable quickly: + +`Conversation → Tool Trace → Deterministic Findings → Release Verdict` + +## Merge Rush media policy + +The public case study may describe verified QA states from the private development record, but it must not invent gameplay screenshots. + +When real public screenshots or video are approved, add them to the public portfolio `assets/` folder and reference them from the case study. Keep the private repository private. + +## Request-form truth + +The current direct request endpoint uses Google Apps Script with `no-cors` browser submission. The browser cannot verify server-side receipt. + +The UI may show a **browser submission reference**, but must not call it a server confirmation until the endpoint returns a verifiable response with appropriate CORS handling. + +## QA + +The Site Preflight workflow remains the release safety net and now includes JavaScript syntax validation for all root JS files. V2 adds `labs.html` and `now.html` to Pa11y and Lighthouse coverage. + +Before merge: + +1. JavaScript syntax validation +2. HTML validation +3. spelling +4. mobile Pa11y +5. Lighthouse +6. broken-link scan +7. interaction review for Recruiter Mode V2, language switch, Ajoop, role deep links and SINAMA Evidence Explorer + +## Migration note + +`flagship-copy.js` was removed. Its responsibilities are now handled by the registry + V2 runtime. + +`legacy-script.js` is compatibility code, not the preferred home for new product facts. New current-state logic should enter through the registry or focused V2 modules first. diff --git a/README.md b/README.md index 99a583e..63e0d07 100644 --- a/README.md +++ b/README.md @@ -2,53 +2,112 @@ Source code for [kaanbalci.com](https://kaanbalci.com), the professional portfolio of **Kaan Balcı — AI Designer & Software Developer**. -## Professional focus +## Current positioning -- Conversational AI and chatbot flow design -- Solution engineering -- LLM response, reasoning, code and multimodal evaluation -- Workflow automation and n8n-style logic -- User-centered software products -- Python, C#/.NET, JavaScript, PHP and SQL +The public portfolio is centered on: -Game development, mobile development and interactive projects remain part of the wider portfolio, while the main public positioning is centered on AI and software roles. +- Applied AI and AI Agent Reliability +- AI Solutions / Solution Engineering +- Conversational AI and workflow design +- LLM evaluation and evidence-based QA +- Python / FastAPI / PostgreSQL product work +- TypeScript product and interactive systems + +Two current flagship products lead the portfolio: + +- **SINAMA — AI Agent Reliability Lab** +- **Merge Rush: Tiny Factory** + +## Portfolio Architecture V2 + +Current product truth is centralized in `portfolio-data.js` and consumed by `portfolio-v2.js`. + +This registry powers or synchronizes: + +- Recruiter Mode V2 +- Role-specific recruiter deep links +- Ajoop evidence answers +- Build Log +- Kaan Labs +- SINAMA Evidence Explorer +- Shared bilingual V2 copy + +See [`PORTFOLIO_ARCHITECTURE.md`](PORTFOLIO_ARCHITECTURE.md) before adding new flagship facts or recruiter evidence. + +## Runtime boot sequence + +`script.js` is now a small compatibility bootloader rather than the monolithic application runtime. + +On pages already migrated to V2 it keeps parser order intact and loads the legacy runtime between the registry and the explicit V2 runtime: + +`portfolio-data.js → script.js bootloader → legacy-script.js → portfolio-v2.js` + +On older pages that still include only `script.js`, the bootloader injects the full stack automatically: + +`portfolio-data.js → legacy-script.js → portfolio-v2.js` + +It also injects `portfolio-v2.css` when a legacy page does not already include it. This keeps Recruiter Mode, Ajoop and current portfolio evidence synchronized across older case studies, certificates, 404 and mini-game pages without rewriting every HTML file at once. ## Main pages -- `index.html` — Landing page, focus areas, recent experience, selected work and contact hub -- `works.html` — Filterable project catalog -- `games.html` — Interactive web games catalog -- `adventure.html` — Career Adventure mini game -- `joyday-paint.html` — Joyday Action Painting experience -- `ai-flow-puzzle.html` — n8n-inspired chatbot workflow puzzle -- `project-detail.html` — Dynamic project detail page -- `blog.html` — Professional experience timeline -- `single-work.html` — Training and course certificate gallery -- `about.html` — Professional profile and capability map -- `request.html` — Project request form -- `script.js` — Navigation, EN/TR translation, Ajoop, Recruiter Mode, Command Palette and dynamic project content -- `style.css` — Responsive site styling +- `index.html` — concise landing page led by flagship evidence +- `works.html` — curated professional project catalog +- `sinama-case-study.html` — Applied AI / reliability flagship case study + evidence explorer +- `merge-rush-case-study.html` — game / interactive flagship case study + QA evidence +- `blog.html` — professional experience timeline +- `about.html` — profile and capability map +- `request.html` — structured project inquiry +- `now.html` — living product / engineering build log +- `labs.html` — technical experiments separated from the main professional narrative +- `games.html` — active game product + playable browser work +- `single-work.html` — training and certificates +- `project-detail.html` — dynamic archive project detail route (`noindex`) + +## Core files + +- `portfolio-data.js` — source of truth for current flagship/project/recruiter/build data +- `portfolio-v2.js` — V2 runtime and evidence surfaces +- `portfolio-v2.css` — V2 component styling +- `script.js` — global compatibility bootloader +- `legacy-script.js` — preserved pre-V2 global runtime: navigation, theme, historical translations, dynamic archive details, chatbot shell, command palette, request/game utilities and other established interactions +- `style.css` — global styling primitives and legacy component styles + +`flagship-copy.js` has been retired; its responsibilities moved to the registry + V2 runtime. + +## Recruiter deep links + +- `/?role=applied-ai` +- `/?role=solution-engineering` +- `/?role=software` +- `/?role=game` + +A valid role parameter selects the matching evidence profile and opens Recruiter Mode. + +## Ajoop scope + +Ajoop remains deterministic and portfolio-focused. V2 grounds its current SINAMA, Merge Rush, role-fit and build-status answers in the central registry. It is **not** presented as a live LLM/RAG assistant in this release. ## Resume -All public resume actions use one centralized resume URL in `script.js`: +The public resume URL is mirrored in the registry and in the preserved legacy runtime: -```js -const resumeLink = "https://drive.google.com/file/d/1eERVaYoP-ICuP3xfbzpaDaCo5amwqA8u/view?usp=sharing"; -``` +`https://drive.google.com/file/d/1eERVaYoP-ICuP3xfbzpaDaCo5amwqA8u/view?usp=sharing` -Public buttons use labels such as **View Resume** or **Download Resume**. Role-specific application CVs are intentionally not exposed on the website. +Role-specific private application CVs are intentionally not exposed. -## Current links +## QA -- Portfolio: [kaanbalci.com](https://kaanbalci.com) -- Resume: [View Resume](https://drive.google.com/file/d/1eERVaYoP-ICuP3xfbzpaDaCo5amwqA8u/view?usp=sharing) -- LinkedIn: [linkedin.com/in/balcikaan](https://www.linkedin.com/in/balcikaan/) -- GitHub: [github.com/UAJOP](https://github.com/UAJOP) -- Email: [kaanb8776@gmail.com](mailto:kaanb8776@gmail.com) +GitHub Actions Site Preflight checks: + +- JavaScript syntax for every root `.js` file +- HTML validation +- spelling +- mobile WCAG 2 AA with Pa11y +- Lighthouse performance / accessibility / best practices / SEO +- broken links + +V2 adds `labs.html` and `now.html` to automated accessibility and Lighthouse coverage. ## Deployment The repository is configured for GitHub Pages with the custom domain in `CNAME`. - -Keep the existing `assets` directory unchanged when replacing the updated code files, because the HTML pages reference the current portfolio images, certificates and logos from that directory. diff --git a/about.html b/about.html index f68da1d..4d8df38 100644 --- a/about.html +++ b/about.html @@ -1,230 +1,11 @@ - - -
- - - - - - - - - - -
-Kaan Balcı
-Open for work
-
-About
-- I am building a career at the intersection of artificial intelligence, automation and software development. - My strength is combining technical execution with practical business understanding. -
-
-Professional profile
-I am an AI Designer and Software Developer focused on conversational AI, workflow automation, LLM evaluation, AI agent reliability and user-centered software products. My background combines software development with AI flow design, product thinking, testing and practical digital operations.
-At CBOT, I worked on enterprise conversational AI projects involving chatbot QA, flow restructuring, multi-channel automation and an insurance claims intake POC. My work included testing, stabilization, channel configuration, voice and WhatsApp transitions, document collection and multi-step workflow logic.
-Through project-based AI training work, I evaluated LLM responses, reasoning, code and multimodal outputs. As a co-founder of Atölye Joyday, I also built a live website, package-based reservation journey and automated operational workflows for a real customer-facing business.
-In 2026, I expanded this direction with SINAMA, a live AI Agent Reliability Lab for deterministic multi-turn workflow evaluation, regression evidence and release readiness, while also developing Merge Rush: Tiny Factory as a Phaser 3 / TypeScript game product with progression, multi-cell board logic and a platform-aware YouTube Playables direction.
-I am currently interested in Applied AI, AI Solutions, Forward Deployed Engineering, Conversational AI, Solution Engineering, LLM Evaluation, Workflow Automation and software development opportunities.
-Milestone journey
-My background combines field operations, software projects, AI workflow design and real business product experience.
-Built practical experience in logistics, on-site coordination, team supervision and time-sensitive execution.
Built university and personal projects across Java, Python, web interfaces and databases.
Focused on C#, MSSQL, Python data analysis and Unreal Engine / Unity prototypes.
Worked on chatbot flows, AI Flow logic, IVR awareness and LLM response quality evaluation.
Built SINAMA as a live agent-reliability product, continued Joyday’s digital workflows and developed Merge Rush as an active Phaser / TypeScript game product.
Capability map
-- I can work across AI product logic, reliability testing, automation design, backend implementation, web interfaces and software quality workflows. -
-How I work
-- Whether it is an AI agent workflow, a dashboard, a website or a game mechanic, my process is to understand the user, - structure the logic, build the system and refine it through testing. -
-Clarify the user need, business requirement, data source, risks and expected result.
-Map the flow, system structure, edge cases, evidence and core interaction path.
-Implement, validate, compare regressions, refine and keep the output maintainable for real usage.
-Toolbox
-Python, C#, JavaScript, TypeScript, PHP, Java, Kotlin, C++
-Conversational AI, AI Flow, n8n-style workflow logic, deterministic agent evaluation, LLM evaluation, prompt analysis and output review
-FastAPI, REST APIs, PostgreSQL, MySQL, MSSQL, SQLite, Firebase, Google Sheets workflows
-Phaser 3, Unity, Unreal Engine, Blueprints, responsive game UI, gameplay logic and platform abstraction
-
Kaan BalcıOpen for workAbout
I am building a career at the intersection of Applied AI, solution engineering and hands-on software development.

Professional profile
My current direction combines conversational AI, AI agent reliability, workflow automation, LLM evaluation and product-minded software engineering. I care about what the system actually does, not only whether the interface looks convincing.
At CBOT, I worked on enterprise conversational-AI projects involving chatbot QA, stabilization, channel configuration, flow restructuring and an insurance-claims intake POC. Project-based AI evaluation work added structured review of LLM responses, reasoning, code and multimodal outputs.
In 2026, SINAMA became my strongest Applied AI / reliability-engineering product, while Merge Rush: Tiny Factory became the main game-system and interactive-product engineering track. Atölye Joyday remains a real-business example of customer-facing product ownership and automation.
Capability map
Milestone journey
Field coordination, logistics and team execution in time-sensitive environments.
Java, Python, web interfaces, databases and university projects.
C#/.NET, SQL, Python data work and Unity / Unreal prototypes.
Chatbot QA, workflow restructuring, multi-channel logic and LLM evaluation.
SINAMA, Merge Rush and continued real-business digital product work.
How I work
Clarify the user, business rule, integration boundary, risk and expected output.
Map workflows, prerequisites, edge cases, state and evidence before hiding complexity behind UI.
Implement, test, compare regressions and keep the system maintainable enough to iterate.
Toolbox
Python, C#, JavaScript, TypeScript, PHP, Java, Kotlin, C++
AI agent evaluation, conversational AI, LLM evaluation, workflow automation, AI Flow, n8n-style logic
FastAPI, REST APIs, PostgreSQL, MySQL, MSSQL, SQLite, Firebase
Phaser 3, Unity, Unreal Engine, Canvas, responsive gameplay and platform abstraction
Choose the evidence view
-Kaan Balcı
-Open for work
-
-Experience
-- I have worked across AI workflows, software development, event operations and entrepreneurship. - This combination helps me understand both technical systems and real business processes. -
-Co-founded a creative workshop studio and developed its digital product and operational experience. Built the responsive website, package-based reservation journey, application forms and automated notification workflows using HTML, CSS, JavaScript, Google Forms, Google Sheets and Apps Script.
- -Open Website -Worked on enterprise conversational AI projects across banking, insurance, municipal services and employee benefits. Contributed to live-chat QA, chatbot stabilization, channel configuration, large-scale flow restructuring and multi-channel automation processes. Designed, built, tested and presented an insurance claims intake POC involving voice interaction, phone verification, WhatsApp handoff, photo and document collection, and workflow coordination.
- -Company Website -Evaluated LLM responses, reasoning, code outputs and multimodal interactions through structured AI training and quality-assessment tasks, focusing on correctness, clarity, instruction following and user relevance.
- -Company Website -Developed a web-based event-management dashboard using PHP, JavaScript, HTML, CSS and MySQL. Built a database-backed interface for event creation, editing, tracking and reporting during a software development internship.
- -Company Website -Coordinated on-site event operations, team supervision, logistics, customer experience and time-sensitive execution for congresses, seminars, corporate meetings and product launches.
- -Supported event logistics, on-site coordination, task distribution and team supervision in fast-paced corporate event environments.
- -Company Website -Activities & Industry Events
-These activities support my technical development but are separate from the professional experience timeline.
-Led a four-person team during a 24-hour competition, supporting game mechanics, Git workflows and rapid Unreal Engine prototyping. The team placed 4th among 15 teams.
-Attended game-industry sessions focused on development tools, game economy and studio operations, expanding practical industry awareness and professional connections.
-
Kaan BalcıOpen for workExperience
The useful pattern across these roles is translating real operational needs into clearer workflows, software or evaluation evidence.
Co-founded a creative workshop studio and built its customer-facing digital product experience: responsive website, package-based reservation journey, application forms and automated notification workflows using HTML, CSS, JavaScript, Google Forms, Google Sheets and Apps Script.
View Case StudyWorked on enterprise conversational-AI projects across banking, insurance, municipal services and employee benefits. Contributed to live-chat QA, chatbot stabilization, channel configuration, large-scale flow restructuring and multi-channel automation. Designed, built, tested and presented an insurance-claims intake POC involving voice interaction, phone verification, WhatsApp handoff, photo/document collection and workflow coordination.
View Portfolio EvidenceEvaluated LLM responses, reasoning, code outputs and multimodal interactions through structured quality-assessment tasks focused on correctness, clarity, instruction following and user relevance.
Developed a web-based event-management dashboard using PHP, JavaScript, HTML, CSS and MySQL, including database-backed event creation, editing, tracking and reporting.
Coordinated on-site operations, team supervision, logistics, customer experience and time-sensitive execution for congresses, seminars, meetings and product launches.
Supported event logistics, on-site coordination, task distribution and team supervision in fast-paced corporate event environments.
Activities & industry
Led a four-person team during a 24-hour competition, supporting game mechanics, Git workflows and rapid Unreal Engine prototyping. The team placed 4th among 15 teams.
Attended game-industry sessions on development tools, game economy and studio operations, expanding practical industry awareness.
SINAMA and Merge Rush now extend the timeline with current evidence in reliability engineering, backend architecture, gameplay systems and product iteration.
Role mapping
Kaan Balcı
-Open for work
-
-Games & Interactive
-The game side of my portfolio now has two layers: lightweight browser experiments and Merge Rush: Tiny Factory, an active Phaser 3 + TypeScript product focused on progression, responsive gameplay and YouTube Playables architecture.
- -Game catalog
Merge Rush shows deeper game-system and platform work. The smaller browser games show interaction design, workflow thinking and rapid prototyping.
A factory-themed merge game with timed orders, restoration progression, multi-cell pieces, staged board unlocks, deadlock-aware logic, Endless mode and platform-aware lifecycle architecture.
View Case Study
Build n8n-inspired chatbot workflows by placing trigger, intent, response, fallback and automation nodes on a visual board.

Choose a square, circle or rectangle canvas, pick colors and use virtual action painting tools to create a downloadable PNG artwork.
Play Game
Combine learning objects, software tools, AI workflow experience and portfolio proof until the final Job Offer object appears.
Play Game
A planned quick decision game about answering interview questions, managing time and choosing the strongest response path.
Future expansion idea — not in active production.Current direction
Merge Rush is the current game-development focus. New mini-game ideas remain secondary while the core product is refined through playtesting, pacing work and production architecture.
Kaan BalcıOpen for workGames & Interactive
Merge Rush is the current game-development priority. Smaller browser experiments remain playable, while technical experiments that are not really games now live under Kaan Labs.
Game catalog
Timed orders, factory restoration, multi-cell board logic, responsive layouts and platform-aware architecture.
View Case Study
Build n8n-inspired chatbot workflows by placing trigger, intent, response, fallback and automation nodes.

Create downloadable action-painting artwork with multiple canvas shapes and virtual tools.
Play
Combine learning, tools, AI workflow experience and portfolio proof until the Job Offer object appears.
PlayCurrent direction
The old Interview Run concept is no longer presented as active portfolio work. Future small experiments belong in Labs until they earn a place in the game catalog.
AI Designer & Software Developer
-My work connects conversational AI, agent reliability, solution engineering, LLM evaluation, workflow automation and product development. I focus on systems that can be tested, explained and improved — not just demos that look good once.
- -My current direction combines Applied AI, agent reliability, solution engineering, conversational AI and hands-on software development. I focus on systems that can be tested, explained and improved.
+ +

What I do
I combine software logic, automation thinking and practical AI experience to design systems that are clear, usable and maintainable.
Multi-turn scenarios, deterministic workflow checks, regression evidence, release-readiness logic and advisory semantic evaluation.
Chatbot flows, intent and response logic, QA, fallback paths, IVR and multi-channel interaction design.
Python/FastAPI services, TypeScript interfaces, C#/.NET systems, APIs, persistence and product-focused frontend work.
Responsive browser games, gameplay state, progression systems, input design, platform adapters and product iteration.
Flagship work
SINAMA shows Applied AI and reliability engineering. Merge Rush shows interactive product, gameplay-state and platform architecture.
Repeatable multi-turn testing, deterministic workflow evidence, regression comparison, version trends and release-readiness decisions across insurance and e-commerce.
A Phaser 3 + TypeScript merge game with timed orders, factory restoration, progressive board unlocks, multi-cell pieces and platform-aware architecture.
Tech stack matrix
A quick matrix showing how technologies fit into real project work rather than a flat keyword list.
Scenario design, agent adapters, deterministic evaluation, regression comparison, semantic shadow evaluation and release evidence.
Automation systems, APIs, database-backed applications and maintainable backend logic.
Responsive products, customer journeys, dashboards, forms and user-facing application flows.
Gameplay architecture, progression, responsive UI, platform lifecycle and browser-based interactive experiences.
Currently building
One focuses on reliable AI systems; the other on product-minded game engineering. Both are being iterated with tests, architecture and real product constraints.
Live reliability product with insurance + e-commerce collections, regression comparison, version trends, release readiness and external agent testing. Current focus: calibration and stabilization.
Phaser 3 + TypeScript merge game with timed orders, factory restoration, staged board unlocks, multi-cell pieces and a YouTube Playables-oriented platform architecture.
Planning dashboards around reservation data, package demand, customer flow and operational tracking.
What I do
Multi-turn scenarios, tool evidence, regression analysis and release-readiness logic.
Translate customer and business requirements into clear workflows, integrations and implementation paths.
Python/FastAPI services, TypeScript products, databases, testing and maintainable product architecture.
Browser games, gameplay state, progression, responsive UI and platform abstraction.
AI workflow demo
Select a scenario and the page will generate a simple AI flow map. This lightweight demo shows how I structure intents, steps and user journeys.
Algorithmic 3D lab
A lightweight canvas demo built with vanilla JavaScript. The shape is generated from a triangular parametric mesh and rendered with perspective projection, depth sorting and mouse-controlled rotation.
Vanilla JS
The mesh is calculated with barycentric coordinates, animated through a wave function and projected onto the canvas without any external 3D library.
Recent experience
My recent work is centered around AI-powered chatbot systems, LLM evaluation, automation flows and real business products.
See ExperienceBuilding the studio’s website, reservation journey, automation workflows and digital product experience.
Contributed to enterprise chatbot QA, flow restructuring, multi-channel automation and an insurance claims intake POC.
Evaluated LLM responses, reasoning, code and multimodal outputs through structured quality-assessment tasks.
Featured AI case study
A Turkish-first reliability platform that runs repeatable multi-turn scenarios, captures tool traces, evaluates workflow contracts, compares agent versions and turns the evidence into release-readiness decisions.

Experience
Recent work spans enterprise conversational AI at CBOT, structured LLM evaluation, and a live customer-facing product at Atölye Joyday.
See ExperienceCustomer-facing digital product, reservation journey and operational automation.
Enterprise chatbot QA, stabilization, channel configuration and workflow restructuring.
Structured evaluation of LLM responses, reasoning, code and multimodal outputs.
Featured works
SINAMA demonstrates AI reliability engineering; Merge Rush demonstrates interactive product and game-system engineering. The rest of the catalog shows the breadth behind those directions.

Live AI reliability product with deterministic workflow evaluation, baseline/regression comparison, version trends, external agent testing and release-readiness evidence.
Active-development merge game connecting timed orders, factory restoration, board-space pressure, progression and platform-ready game architecture.
View Case Study
A live n8n-inspired browser game demonstrating chatbot flow structure, fallback logic, validation and user journey thinking.

A live digital product combining package-based reservation journeys, automated notifications and operational workflows.

A database-backed C# Windows Forms project demonstrating patient, doctor, secretary and appointment workflows.
Selected evidence
Enterprise conversational-AI workflow, QA and stabilization evidence.
View DetailsReal-business product ownership: web UX, package flow, reservations and automation.
View Case StudyC#/.NET and SQL Server workflow project preserved as a source archive.
View Case StudyMore experiments on GitHub
The main catalog highlights selected projects, while GitHub includes more university work, prototypes and experiments.
University and OOP practice repositories.
Automation, data and learning projects.
Frontend logic and UI practice.
JavaScript API and interface practice.
Latest build
Contact hub
I am especially interested in Applied AI, AI Solutions, Forward Deployed / Solution Engineering, Conversational AI and product-minded software roles.
Contact
Use Recruiter Mode for role-specific evidence, or open the full project catalog.