Radar 30X
An AI-assisted social media growth tool that gathers public topic signals, ranks them with explicit code-based weights and drafts audience-specific content ideas. It connects audience research, trend analysis and content creation, with configurable topics such as AI and fitness.
The product, shown without personal or sensitive data
What is implemented today
- Collect public topic signals with source-specific coverage limits.
- Rank signals through explicit, configurable code-based weights.
- Assist audience-fit classification and draft carousel, reel and LinkedIn ideas.
- Keep proposed future topics distinct from validated predictions.
Context
Growth work needs a way to connect what is being discussed publicly with a specific audience, instead of collecting disconnected popular posts.
Problem
Signals from different networks are incomplete and not directly comparable. An AI-written trend list can hide source gaps, ranking assumptions and the distinction between observed activity and an editorial suggestion.
My role
Product and workflow design, source integration, ranking logic and AI-assisted application development.
What I built
A public web MVP with topic configuration and result views, a Python collection and scoring pipeline, typed JEV classification, DeepSeek-assisted drafting, public search through Tavily and cloud execution through GitHub Actions. Code determines the ranking; language models assist classification and writing.
Progress achieved
The MVP connects source signals, visible ranking assumptions and audience-specific drafts in one workflow. Business adoption, incremental reach and predictive performance have not been established by this portfolio case.
Data or evidence used
The public application and repository document connectors for APIs, public RSS and indexed web search. Available growth metrics depend on the source; search presence is not treated as equivalent to verified platform engagement.
Limitations
Coverage depends on source access, indexing, quotas and configuration. Missing network metrics limit comparisons; generated explanations can be wrong. Proposed future trends require prospective evaluation. The portfolio review did not launch paid runs or independently audit the reported per-run cost.
Next validation step
Evaluate ranking usefulness with a defined audience, log source coverage and measure whether reviewed content experiments improve outcomes against a baseline.