Seekho Platform
AI Product Strategy & Technical Product Management
A local educator platform that moved from a Streamlit demand-validation MVP to an offline-first, async curriculum engine with reliable AI delivery.
By Shehryar Hassan·2025-11-01
Next.jsTypeScriptSupabaseInngestStreamlitPythonLLMs
seekhopunjab.vercel.app
The Product VisionScaling a localized educational generator from a low-fidelity validation prototype into a production-grade learning platform.
The User Problem
Educators needed custom learning materials mapped to regional textbook boards, but the idea first needed demand validation. Once validated, large generative lesson plans created a second problem: long compilation loops froze the interface and caused standard API gateways to drop requests.
Architecture
- Shipped a rapid Streamlit MVP with SQLite content parsing to test curriculum-matching demand before investing in the full platform.
- Rebuilt the validated loop in Next.js and Supabase, then moved long generation work into an asynchronous, event-driven pipeline using Inngest background workers.
- Persisted intermediate state in Supabase so users can close the browser mid-run and return to a completed plan later.
Product & Growth Impact
- Validated the core product hypothesis in days and used the interaction signal to justify scaling from MVP to v2.
- Reduced processing failures to 0% by shifting heavy computation off the request thread.
- Kept the UI fully interactive during long AI workloads, lifting retention through generation.