AI Engineer

AI is not a feature we bolt on; it is the engine of everything we ship. DijiPilot generates complete brands, product lines, and storefront content through multi-step pipelines that orchestrate OpenAI, Anthropic, Gemini, and a dozen image generation providers. DijiFlow turns speech into text entirely on-device, built on Whisper with custom vocabulary correction and hallucination filtering across 29 languages.

You will own these systems end to end: designing pipelines and prompts, choosing and integrating models, building evaluation harnesses, and turning research ideas into features used daily around the world. Like everyone at PITIR TECH, you will spend most of your working day with AI, and with AI agents doing the heavy lifting alongside you.

What you will do

• Design, build, and tune the multi-step AI pipelines that generate brands, products, and content for DijiPilot.
• Integrate and benchmark LLM and image generation providers behind our provider abstraction, and decide which model wins each job.
• Improve DijiFlow’s on-device speech recognition: vocabulary correction, hallucination filtering, and per-language quality.
• Build agentic workflows and automations that multiply the whole team’s output, not just your own.
• Measure quality with real evaluation harnesses; never guess.

What we look for

• Real engineering depth: you are a software engineer first, with strong TypeScript or Python.
• Production experience with LLM APIs, prompt design, and multi-step AI workflows or agents.
• You already work AI-first: coding with AI assistants, building automations, and delegating to agents is your normal day.
• A rigorous, measurement-driven approach to model output quality.
• Comfortable working in English.

Nice to have

• On-device inference: Whisper, CoreML, Metal, CUDA, or GGML.
• Image generation workflows and asset post-processing.
• Node.js backends, PostgreSQL, AWS.
• Turkish language skills.