Data Infrastructure for Amazon Seller Central

Description: We are an e-commerce brand selling products on Amazon and would like to build a reliable internal data infrastructure for our Amazon Seller Central and Amazon Advertising data. The goal is to automatically access data from the Amazon Selling Partner API and Amazon Ads API, store it in a structured database, and make the data available for reporting, dashboards, and later access via an LLM such as ChatGPT or Claude. Project Scope: We are looking for an experienced freelancer who can help us design and implement the following setup: Amazon API Integration Connect to Amazon Selling Partner API Connect to Amazon Ads API Set up secure authentication and credential handling Pull historical data and enable recurring updates Handle API limits, async reports, retries, and error logging Data Sources We would like to access and store data such as: Sales and order performance SKU / ASIN / parent ASIN data Business reports such as sessions, page views, conversion rate, buy box percentage Inventory data Returns data, if available Settlement / fee data, if available Sponsored Products campaign performance Sponsored Brands / Sponsored Display data, if relevant Search term reports Campaign, ad group, targeting, placement, spend, sales, orders, ACOS, ROAS, CPC, CTR, CVR Database Setup Recommend and set up a suitable database architecture We are open to PostgreSQL, BigQuery, Snowflake, or another suitable solution Create a clean raw-data layer and analytics-ready tables Build a clear schema for daily and monthly reporting Include mapping tables for SKU, ASIN, product category, parent product, campaign, portfolio, and advertising intent Data Transformation & KPI Logic Normalize API data into clean tables Create views or tables for common business KPIs Examples: Sales, Units, Sessions, Conversion Rate, Ad Spend, Ad Sales, ACOS, TACOS, ROAS, CPC, CTR, CVR, Returns, Fees, Contribution Margin if possible Enable monthly, YoY, MoM, and rolling 3-month analysis Document all KPI definitions clearly Automation Set up scheduled data refreshes Ideally daily updates with rolling refresh windows Store historical data reliably Include monitoring or alerts for failed jobs Provide clear documentation so the setup can be maintained later LLM Access Layer In a later step, we want to be able to ask questions about the data using an LLM such as ChatGPT or Claude. Examples: “Which products had the strongest ACOS increase last month?” “Which campaigns have high spend but declining sales?” “Which search terms should be added as exact keywords?” “Which products are losing sales YoY despite increasing sessions?” We need guidance and implementation for a safe read-only setup where an LLM can query approved database views without accessing sensitive raw data or API credentials. Preferred Experience: Strong experience with Amazon Selling Partner API Strong experience with Amazon Ads API Experience with Python-based ETL/ELT pipelines Experience with PostgreSQL, BigQuery, Snowflake, or similar databases Experience with data modeling for e-commerce / Amazon marketplace data Experience with dbt, Airflow, Prefect, Dagster, or similar tools is a plus Experience with LLM SQL agents, LangChain, LlamaIndex, OpenAI API, or Claude API is a plus Good documentation skills Deliverables: Working Amazon SP-API and Amazon Ads API connection Automated data import pipeline Structured database with raw and analytics-ready tables Initial historical data backfill Scheduled refresh process KPI views / reporting tables Documentation of setup, schema, credentials handling, and update logic Recommendation for how to securely connect ChatGPT or Claude to the data later Important: We are not looking for a one-time manual export solution. We want a scalable, maintainable setup that can serve as the foundation for ongoing Amazon performance analysis and AI-supported decision making. Please include in your application: Relevant experience with Amazon SP-API and Amazon Ads API Examples of similar data pipeline or analytics projects Your recommended tech stack for this project How you would approach the first MVP Estimated timeline and budget range Any questions you would need answered before starting