Stream JSON Repair Interceptor
Self-healing parser resolving unescaped keys or trailing commas mid-stream.
const cleanJSON = (str) => { return str.replace(/\,\s*\}/g, '\u0027}\u0027').replace(/\,\s*\]/g, '\u0027]\u0027'); };Conversational E-Governance AI
Bilingual Citizens AI Assistant
Developed a state-level bilingual chatbot with Next.js, Google GenAI SDK, and custom Nastaliq CSS, solving stream parsing failures and dynamically generating stateful forms in the viewport.
- Sector
- Conversational E-Governance AI
- Timeline
- 5 months
- Role
- Lead AI Architect
- Next.js
- Prisma
- PostgreSQL
- Google GenAI SDK
- Gemini 2.5
- Urdu Nastaliq CSS
- Instant
- Form Ingestion Time
- 100%
- JSON Parsing Success
Self-prefilling known profiles dynamically
Ensured by self-healing stream interceptors
The problem
Ensuring structural JSON schema output consistency during multi-turn chats under high API traffic, and creating dynamic user intakes without static layouts.
The solution
Formulated a server-side JSON-repair interceptor and designed a stateful chatbot pre-filling user files and injecting custom intake inputs directly into stream contexts.
Approach
How it was built.
01
Structured Parsing Safeties
Implemented server-side parser middleware that monitors incomplete stream packets and repairs brackets, maintaining absolute JSON compliance.
02
Stateful Form Injection
Programmatically mapped profile data and dynamically evaluated fields inside the conversation context, generating specialized forms inside chat viewports.
Technical detail
The parts worth explaining.
What I took from it
- Stream validators must handle packet fragmentation
- Nastaliq typography requires specialized CSS container constraints
Where it goes next
- Deploy native text-to-speech outputs in localized regional Urdu dialects
- Integrate government citizen portal SSO APIs
Got a problem shaped like this?
If something here looks like what you're facing, tell me about it. The first conversation is an architecture discussion, not a sales call.