2026Oğuz Kaan YIılmaz
AI production team — Istanbul & remote
Zero-Shot Team is an independent group of self-taught AI builders based in Istanbul. It builds AI agents, enterprise retrieval (RAG), fine-tuning and evaluation pipelines, vision and voice systems, generative design systems and LLM infrastructure. Every member runs their own profile and portfolio, and clients keep the code, prompts and evaluation sets.
Everyone has a page and edits it themselves. Sigils are generated from the handle — computed, not drawn.
0106
No consulting deck. Working systems, with numbers, at a known price.
The numbers are the clients' own measurements.
The only entry requirement here.
Nobody asks where you studied. Show something that runs and we'll talk.
Every member publishes what they're learning right now. Not knowing isn't embarrassing; hiding it is.
Every project ships with an eval set. That's the line between an impressive demo and a system.
Code, prompts, datasets and eval files are part of the handover. No lock-in.
The team is open. Make an account, set up your page, fill in your portfolio. We review every two weeks.
Notes on decisions we ran into on real projects — what we actually looked at when we had to choose.
The things people ask about Zero-Shot Team.
Zero-Shot is not a software agency or a consultancy; it is a builder team that assembles per project. It does not sell an off-the-shelf SaaS product or advertising services.
Zero-Shot Team is an independent group of self-taught AI builders based in Istanbul. It builds AI agents, enterprise retrieval (RAG), fine-tuning and evaluation pipelines, vision and voice systems, generative design systems and LLM infrastructure.
In machine learning, Zero-Shot describes a model doing a new task without having seen any examples of it. The team's name comes from there: none of the members studied this formally, everyone taught themselves. The brand name is Zero-Shot Team.
The team is made up of self-taught agent architects, fine-tuners, retrieval engineers and generative designers. Every member has their own profile page and portfolio and runs it themselves. The team is open, and reviews happen every two weeks.
Six areas of work: AI agents and automation, RAG and enterprise search, fine-tuning and evaluation, vision and voice systems, generative design, and LLM infrastructure and cost. Every project ships with an evaluation set and a written definition of done.
You describe what you want to build and within two days we tell you who takes it. The first step is always the same: writing down how the work will be measured. Development does not start before the evaluation set exists.
You do. Code, prompts, datasets and evaluation files are part of the delivery. No lock-in and no single-vendor dependency — the system is handed over in a state another team could maintain.
The team is based in Istanbul and works remotely. We work in Turkish and English, and take work from inside and outside Türkiye.
Tell us what you want built and we'll tell you who's picking it up within two days.