Role overview
**Zzazz**
is a
**US-based AI company**
transforming how digital content is valued and monetized. Our breakthrough
**Large Pricing Model (LPM)**
turns content into a real-time tradable asset — dynamically assigning accurate market values based on billions of engagement signals, live user interactions, and real-time market data.
***Role: Own prompts and evaluation for AI that generates reliable React Native layouts/components—ideal for a data‑science‑leaning engineer who can read and reason about frontend code.***
**Responsibilities**
* Design, test, and version prompts for React Native UI generation (components, props, styles, Flexbox).
* Build offline/online evaluation: golden sets, JSON/schema checks, visual/snapshot tests, and A/B experiments with clear metrics.
* Analyze logs and failure clusters; curate/label datasets and improve prompt chains, function/tool calling, and retrieval.
* Implement guardrails (prompt‑injection defenses, PII redaction) and reliability fallbacks.
* Partner with FE to integrate outputs; contribute light TypeScript/Python orchestration and test harnesses.
**Requirements (Must‑have)**
* 3+ years in Data Science/ML or Applied NLP/LLMs, with prompt engineering in production.
* Strong Python (pandas/numpy) and SQL for analysis, metrics, and experiment design (stats/A-B basics).
* LLM fundamentals: sampling controls, embeddings/RAG basics, structured output & function/tool calling.
* React Native literacy: read/modify JS/TS component code; solid Flexbox understanding.
* TypeScript and Git/GitHub; CI/testing (Jest/RTL/Playwright); fluency with JSON/JSON Schema/OpenAPI.
* Clear written communication and product sense.
**Nice to have**
* Expo, design tokens, and Figma‑to‑code workflows.
* Vector DBs/RAG, prompt eval tooling, and tracing/observability.
* Cloud/DevOps basics (Docker, GitHub Actions, Vercel/Netlify).