← InstaLILY Interview Insights
The scraping part ate way more time than I budgeted.
Start by clarifying the scope and constraints (e.g., which e-commerce site, expected query types, and deployment environment) to show you think before coding. Then outline a phased plan: data acquisition, AI agent integration, and full-stack development, emphasizing prioritization and trade-offs to deliver a working MVP within 48 hours. Conclude by discussing how you would validate each component and handle potential failures.
Pro tip: Emphasize building a thin vertical slice first—a minimal end-to-end flow from scraping to frontend—to ensure integration early, then iterate. This demonstrates pragmatic prioritization and risk mitigation, which is highly valued in fast-paced AI engineering roles.
Ask questions to understand the target e-commerce site, expected query complexity, user load, and deployment constraints. This ensures you build the right thing and avoid wasted effort.
Sketch a high-level architecture: scraper, data storage, AI agent (e.g., RAG with LLM), backend API, and frontend. Select tools based on speed of development and reliability (e.g., Python with BeautifulSoup/Scrapy, FastAPI, React, and a vector DB like Pinecone or FAISS).
Write a custom scraper to extract product data, handling pagination and rate limits. Clean and store the data in a structured format (e.g., JSON or database) and generate embeddings for retrieval.
Develop a retrieval-augmented generation (RAG) agent that uses the scraped data to answer queries. Expose it via a REST API (e.g., FastAPI) with endpoints for chat and data retrieval.
Create a simple chat interface (e.g., React) that communicates with the backend. Test the end-to-end flow, handle errors gracefully, and deploy if time permits.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.