Available for opportunities

ShubhamKumar
Prajapati

Fullstack Developer & AI/LLM Engineer

I build systems that see, think, and respond in real time — from GPU surveillance pipelines to custom LLM inference stacks.

Open Resume ↗

Projects Shipped

12+

Vision Pipelines

Real-time · GPU

Experience

3+ yrs

Scroll

// tools I build with

// about

Building the
infrastructure
AI runs on.

Shubham Kumar Prajapati

Vadodara, India · @Mostedge

Based in Vadodara, India. For the past 3 years I've built and shipped production systems where AI meets infrastructure — real-time video analytics, event-streaming pipelines, and self-hosted LLM deployments.

Currently at Mostedge, where I design and build a GPU-accelerated computer-vision platform and an in-house AI chat platform.

Featured Work

PRODUCTIONGPUREAL-TIME

GPU Surveillance Platform

Microservice C-store surveillance — face recognition, theft detection, safety analysis. Processing 2× 1080p feeds at 20fps with GPU-accelerated inference.

GStreamer · InsightFace · Kafka · FastAPI · CUDA

LLM INFRAFINE-TUNINGvLLM

Custom AI Chat Platform

In-house AI chat product backed by a custom LLM inference stack — model fine-tuning, context pipelines, and runtime switching across 15+ self-hosted models.

vLLM · Ollama · RunPod · Angular 17 · Tailscale

$ df --help
→ chat   multi-model AI chat
→ build  visual site gen
22 local-first tools loaded

LIVE PRODUCTAI-NATIVEOPEN

DevForge

AI-native developer workbench with multi-model chat, visual site generation, and 22 local-first tools. Live in production.

Next.js · TypeScript · Claude API · Oracle Cloud

Visit devforge.playkashyap.com ↗

ShubhamKumarPrajapati

Building theinfrastructureAI runs on.