Ahmedabad · Generative AI · Open to Roles

Samir
Gajjar.

Full Stack Generative AI Engineer

Building production-ready AI systems and intelligent full-stack applications — RAG pipelines, AI agents, and real-world automation.

Gemini APILangChainQdrantNext.jsPythonMongoDB

AI Engineer

About

AI Engineer × Builder × Entrepreneur × Creative.

Information Technology graduate focused on building real-world AI systems, not experiments. I specialise in production-ready generative AI applications that combine strong backend architecture with intelligent user experiences.

My work spans agentic RAG systems, voice-based AI automation, and full-stack AI platforms. I deploy scalable systems using modern AI infrastructure — vector databases, LLM orchestration frameworks, and real-time streaming interfaces.

“Building AI systems that actually get used.”

AI Systems
RAG, agents, pipelines
Full Stack
Frontend to backend infra
Real World
Production deployments

Core Stack

Generative AI

LLMsRAG SystemsAgentic WorkflowsPrompt EngineeringSemantic Search

Backend & Infra

Node.jsPythonExpressMongoDBFastAPIREST APIs

Frontend

React.jsNext.jsTypeScriptTailwind CSS

AI Stack

Gemini APILangChainQdrantHuggingFace

Work

01Hero Project

Salon Appointment Voice Calling Agent

AI Voice Agent · Automation · Business AI

Conversational voice AI agent that handles bookings, rescheduling, and service queries in real time — simulating a human receptionist. The system understands caller intent, maintains conversational context across multiple turns, and executes backend workflows without human intervention. Built end-to-end: speech-to-text → LLM reasoning → intent execution → text-to-speech response.

Caller→ STT→ LLM Reasoning→ Intent Execution→ TTS→ Response

Voice AIGenerative AISpeech-to-TextTTSMERNReal-time

System Architecture

🎙️Voice Input

↓Speech Recognition (STT)

↓LLM Reasoning Engine

↓Intent Classification

↓Backend Execution

↓Text-to-Speech (TTS)

📞Caller Response

SPT AI Lead Capture & Qualification Bot

Multi-agent · Conversational AI · Lead Intelligence

Multi-agent AI system built for ShreePrasad Technologies. Handles visitor conversations, extracts lead data through natural dialogue, and calculates qualification scores (0–100). Pipeline: Conversation Agent → Qualification Agent → Scoring Agent → Stage Derivation → CRM Agent → Founder Notification. High-intent leads trigger real-time alerts. Built with Gemini API and SSE streaming.

Gemini APIMulti-agentNode.jsSSEMongoDB

ShreePrasad Technologies Website

Full-Stack Web · Agency Platform

Production Next.js App Router website for ShreePrasad Technologies. SSR, dynamic OpenGraph metadata, JSON-LD structured data, Framer Motion animations. Real dual-email transactional workflow via Resend: visitor submits form → agency receives full lead notification → visitor receives branded confirmation email.

Next.jsTypeScriptNode.jsExpressResend

Personal AI Portfolio & Assistant

RAG · AI-Powered Portfolio

Next.jsGemini 2.5 FlashLangChain

This website. Features an embedded AI assistant trained on personal work via RAG. Visitor queries are embedded with HuggingFace sentence transformers, matched against Qdrant vector database, and passed to Gemini 2.5 Flash with grounded context. Session memory, Upstash Redis rate limiting, and streaming responses.

Agentic RAG Chatbot

RAG · Semantic Search · Memory

RAGLangChainGemini

Production-ready RAG system with document ingestion, semantic search, and context-aware reasoning. Supports multi-session memory, intelligent context assembly, and streaming responses grounded in retrieved knowledge chunks.

Steganography Encryption System

Earlier Work

Security · Cryptography · Image Processing

Ask Samir

A custom RAG-powered AI trained on Samir’s projects, experience, and engineering work. Ask anything — like you would ask him directly.

samir.ai — gemini 2.5 flash + qdrant rag

v1.0

How it works

Your question

Embedded using HuggingFace sentence transformers.

Qdrant search

Matched against portfolio knowledge chunks by semantic similarity.

Grounded response

Gemini 2.5 Flash answers using retrieved context only.

Streaming

Response streamed in real time to the interface.

Gemini 2.5 Flash · Qdrant · LangChain · Upstash Redis

Experience

Feb 2026 - May 2026Concluded

Generative AI Intern

Worked on AI-driven applications and real-world generative AI systems. Built production-ready solutions with focus on agentic systems and RAG pipelines.

Generative AILLM IntegrationRAG Systems

Jul 2025

MERN Stack & AI Intern

Softcolon Technologies

Built an agentic RAG chatbot using React, Node.js, LangChain, Qdrant, and Gemini API. Implemented document ingestion pipelines and semantic search capabilities.

ReactNode.jsLangChainQdrant

2022–2026Education · Complete