Sayanta Neogi
Software Engineer • AI & RAG Architect • Cloud DevOps
Experienced Full Stack Developer and AI Engineer with 4+ years of expertise in building scalable web applications, implementing AI-powered solutions, and establishing DevOps practices. Proficient in modern frameworks including React, Laravel, Django, Flask, and cutting-edge AI technologies such as Amazon Bedrock, LangChain, Hugging Face, Model Context Protocol (MCP), and RAG architectures. Skilled in containerization with Docker, CI/CD pipelines, and cloud deployments across AWS, Azure, and GCP.
// sparrow84001/mcp-seo • AI-Powered SEO & Growth Audit MCP Server
import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
import { z } from "zod";
import { discoverProject } from "./analyzer/discovery.ts";
export const server = new McpServer({
name: "mcp-seo",
version: "1.0.4"
});
// Register tool: Framework & Route Discovery
server.registerTool(
"seo_discover_project",
{
description: "Discovers framework (Astro, Next.js, Laravel), routes & metadata.",
inputSchema: { projectPath: z.string().describe("Project root directory") }
},
async ({ projectPath }) => {
const discovery = await discoverProject(projectPath || ".");
return { content: [{ type: "text", text: JSON.stringify(discovery, null, 2) }] };
}
);Live MCP & RAG Playground
Execute live simulations of the authentic Model Context Protocol tools and RAG search pipelines Sayanta engineered. Inspect real AST analysis, framework discovery, and vector embeddings.
Discovers web framework (Astro, Next.js, Laravel), routes & static metadata from project root AST.
{
"projectPath": "d:/SNtemp/astro/SN_portfolio",
"detectSitemap": true,
"scanRoutes": true
}Enterprise AI & AST Pipeline Topology
Interactive schematic of Sayanta’s production architecture connecting LLM agent clients with Model Context Protocol servers, static AST analyzers, and PostgreSQL pgvector RAG stores.
LLM Agent Client
Claude Desktop / Cursor / Web Agent
MCP Server Gateway
sparrow84001/mcp-seo v1.0.4
AST & Static Analyzer
Framework-Aware Parser
Amazon Bedrock
Titan Text v2 Embeddings
PostgreSQL pgvector
HNSW Cosine Vector Index
MCP Server Gateway • sparrow84001/mcp-seo v1.0.4
High-performance TypeScript server built on Bun runtime, exposing Zod-validated auditing and framework discovery tools.
Engineering Case Studies & Repositories
Verifiable research artifacts from Jefferson Lab & BNL alongside production AI architectures and open-source MCP tooling from @sparrow84001.
CaloGAN: Accelerating Detector Simulation 100x via Deep GANs
Brookhaven National Lab (BNL) & Jefferson Lab ATHENA Collaboration
Mentored by Dmitry Romanov (Jefferson Lab) within the Software & Computing Working Group.
Classical Geant4 Monte Carlo simulations model particle showers step-by-step, taking hours per event and creating massive computational bottlenecks for collider physics research.
Generative Adversarial Network (GAN) trained on Jefferson Lab calorimeter energy deposit showers, substituting classical ray-tracing with one-shot neural tensor generation.
mcp-seo: Framework-Aware Model Context Protocol Server
Open-Source AI Tooling Engine on Bun Runtime
Authored and published by Sayanta Neogi as an open-source MCP server for AI coding agents.
AI development assistants lack AST-level static awareness of routing architectures, canonical URL consistency, and framework metadata across Astro, Next.js, and Laravel.
Bun-native Model Context Protocol server exposing Zod-validated audit tools, framework discovery analyzers, and non-destructive unified diff patch generators.
vartualT: Socratic Multi-Agent Intelligent Tutoring Engine
Strict Pedagogical AI Guided by Conceptual Decomposition
Engineered as an autonomous educational platform that avoids direct answer hallucination.
Commercial LLMs frequently dump direct answers to homework problems, destroying critical thinking and failing to detect conceptual prerequisites.
Multi-agent supervisor pattern with dedicated hint-generation, misconception-diagnosis, and Socratic questioning agents orchestrating student guidance.
Enterprise Bedrock + pgvector RAG Pipeline
Sub-Second Semantic Retrieval Over 500k+ Document Chunks
Architected and deployed at Web Mind Ltd with strict enterprise tenant isolation.
High-throughput unstructured knowledge retrieval across diverse enterprise documents while upholding zero-trust data confidentiality agreements.
Amazon Bedrock Titan Text v2 embeddings indexed via PostgreSQL pgvector with HNSW cosine distance (<=>) and VPC endpoint isolation.
mcp-seo
AI-powered SEO and growth audit Model Context Protocol (MCP) server for modern web projects. Analyzes technical, on-page, AEO, GEO, and conversion metrics with framework-aware fixes.
vartualT
Multi-agent AI tutoring platform with strict Socratic methodology — never gives raw answers, only guides students with hints, concept breakdowns, and Correct/Wrong evaluations.
whiteBoard
Interactive digital whiteboard platform built from scratch with custom drawing primitives, real-time collaboration canvas, and export capabilities.
Azure-quantum
Quantum algorithm development and basic simulation scripts utilizing Microsoft Quantum Development Kit and Q# programming language.
CaloGAN
Calorimeter simulation using Generative Adversarial Networks (GANs) for high-energy particle physics research (BNL + Jefferson Lab ATHENA collaboration).
GAN_cryptography
GAN cryptography — neural network generative adversarial architecture applied to encrypted communication, security, and cipher generation.
Professional Experience & Research
4+ years engineering scalable full-stack applications, production LLM systems with Amazon Bedrock & pgvector, and particle detector simulations at national laboratories.
Software Engineer (Full Stack + AI & DevOps)
- Architected enterprise Laravel, React, and Python systems with high throughput and strict data isolation.
- Designed and deployed production RAG pipelines using Amazon Bedrock, pgvector, and LangChain with sub-second retrieval.
- Built autonomous AI agents and document intelligence workflows for PDF, image, and knowledge-base extraction.
- Optimized PostgreSQL and vector database indexes for high-concurrency AI search workloads.
- Engineered scalable cloud architectures across AWS, Azure, and GCP with strict zero-trust confidentiality controls.
Full Stack Web Developer, AI & DevOps Engineer
- Engineered dynamic full-stack web applications utilizing Laravel, Flask, and Django with responsive React frontends.
- Integrated GraphQL and RESTful APIs for high-throughput, low-latency client-server interactions.
- Implemented generative AI features using LangChain and Hugging Face models for NLP and conversational intelligence.
- Built intelligent agent systems and RAG architectures for enhanced search and knowledge retrieval.
- Established automated CI/CD pipelines using GitHub Actions and GitLab CI/CD for containerized Kubernetes deployments.
- Integrated Vite as frontend build tool for faster build times and developer velocity.
Software & Computing Working Group
- Developed calorimeter simulation pipelines using Generative Adversarial Networks (GANs) for high-energy particle physics.
- Collaborated with international scientific research teams under the supervision of Dmitry Romanov (Jefferson Lab).
- Accelerated computational simulations using Python, C++, and deep generative models.
Research Developer
- Utilized the ESCALATE framework and Geant4 for particle physics detector simulations and analysis.
- Developed high-performance simulation software using C++ and Python for particle collision modeling.
- Published code and simulation artifacts under the mentorship of Dmitry Romanov.
Master of Science in Computer Science
Ramakrishna Mission Residential College, University of Calcutta
Bachelor of Science in Computer Science
Kanchrapara College, University of Kalyani
Skills & Technologies
Full-stack web engineering, cloud-native containerized architectures, and specialized AI frameworks including Amazon Bedrock, pgvector, and Model Context Protocol.
AI Engineering & LLMs
Full Stack & Web
Cloud & DevOps
Databases & Storage
Systems & Quantum
Initiate a Conversation
Send an engineering inquiry, collaboration proposal, or project discussion. I review all inquiries directly and typically reply within 24 hours.
Direct Engineering Channels
Reach out directly for AI engineering, full-stack web platforms, or cloud DevOps roles.
Open for senior AI/ML engineering, scalable web architectures, and cloud consulting contracts worldwide.