Collaborator Profile - Beril

Beril Ozbay - Short Bio

Beril Ozbay began her career in city planning and urban design, a background that taught her to think in terms of large systems, complex constraints, and human experience simultaneously. That systems-driven mindset eventually led her into architectural visualisation, where she discovered not only a passion for image-making, but also a fascination with the structured processes behind it. What started with non-destructive workflows, Photoshop actions, and scripting gradually evolved into AI pipelines and tool building. Today, Beril works as AI Lead at Blink Image, where she focuses on integrating generative AI into CGI production workflows while also helping upskill teams across the studio. Alongside her studio work, she is developing an experimental side project called the Design Alignment Agent, exploring how AI might support design decision-making rather than simply generating imagery.

Beril describes herself as “a vibe coder who’s slowly starting to understand what she’s actually doing.” Before AI, her technical experience mostly involved modifying existing scripts rather than writing software from scratch. Now, with AI assistance, she is building and deploying backend systems using technologies such as FastAPI, Vertex AI, Cloud Run, and Firestore. Despite that progression, she remains refreshingly honest about the learning process, describing most of her growth as self-taught through experimentation, documentation, community workshops, and repeatedly breaking and fixing things. She credits much of her success not to technical perfection, but to visual thinking, iterative workflows, and an architectural mindset of “block first, refine later” — building rough working systems quickly before gradually improving them.

For Beril, the biggest advantage of vibe coding is empowerment. AI has dramatically reduced the barrier to entry, allowing creative professionals to prototype and ship tools in days rather than weeks, while staying close to the industries and problems they already understand. At the same time, she is highly aware of the risks. Hallucinated code, unreliable long-context memory, and overconfidence from AI systems can quickly lead projects in the wrong direction if users are not paying attention. Her workflow now intentionally involves cross-checking outputs between multiple models, such as Gemini, ChatGPT, and Claude, in order to spot inconsistencies and reduce errors.

Advice

Her advice to newcomers is grounded in experimentation, structure, and critical thinking:

Ultimately, Beril sees vibe coding as an incredible entry point into software creation, but not the final destination. In her view, AI can help people start building faster than ever before, but long-term success still depends on learning how systems work beneath the surface.