Lead AI Engineer

Lead AI Engineer Full-time • Turkey • Remote / Hybrid in Istanbul About RozyLabs RozyLabs is a Physical AI startup for offline retail. We deploy autonomous robots and kiosks in supermarkets, malls, and airports, focusing on three areas: retail media, Voice AI, and data & insights. Our core product is an AI voice assistant running on kiosks and robots. Visitors can ask for a store, product, service, or destination and get an accurate spoken answer and directions in real time. The Role We're looking for an AI Engineer to improve and scale the intelligence behind the product. You'll work closely with the CTO in a small team and have broad ownership across the voice, retrieval, and backend stack, from speech-in to speech-out. This is a hands-on engineering role. You'll investigate real production failures, build evals around them, improve the system, and measure whether it actually got better. What You'll Work On Real-time voice: Improve speech recognition in noisy, echo-heavy public environments, optimizing accuracy, latency, and reliability across models and providers. RAG and retrieval: Own our retrieval stack across vector, full-text, structured, and hybrid search, including the evals and data models behind it. End-to-end architecture: Improve the speech-to-speech pipeline while managing latency, cost, quality, and reliability as the fleet grows. Venue onboarding: Build and improve the systems that take a new venue from raw data to a reliable production assistant. AI evaluation: Turn production failures into reproducible tests, benchmarks, and experiments. Model and tooling research: Evaluate new LLMs, ASR/TTS providers, agent frameworks, and infrastructure, and adopt them when they meaningfully improve the product. What We're Looking For Minimum 6 years of experience Background in Computer Science or a related engineering field Strong Python and backend engineering fundamentals Experience building and operating production software Experience in a startup or high-tech environment Production experience with LLM applications and RAG systems Ability to diagnose failures across retrieval, prompting, generation, data, and infrastructure rather than treating the LLM as a black box Experience designing or working with structured data models Ability to reason quantitatively about latency, cost, quality, and reliability Ability to turn ambiguous product problems into measurable experiments and evals Regular use of agentic coding tools such as Claude Code, Codex, or similar. AI-assisted development is a core part of how we build. Nice to Have Full-stack engineering experience ASR, TTS, VAD, or real-time audio experience Streaming or low-latency systems Search or information retrieval systems GCP or similar cloud infrastructure Computer vision or camera analytics What We Offer: Competitive salary Equity participation in a fast-growing startup Flexible working model: remote / hybrid in Istanbul Being part of a global expansion story Real-world impact: see your work running live on robots and kiosks in malls, supermarkets, and airports High ownership and direct access to the CTO and founders Full access to the best AI and agentic coding tools