S·R

AI ENGINEER · CHENNAI, INDIA

Shivprasad
Roul

GenAI Backend

Building the intelligent backend layer — from RAG pipelines to agentic systems.

See My WorkRead My Blog
2×Hackathon Winner
10K+Docs Processed
250+Team Led

ABOUT

The Builder
Behind the Code.

I'm an AI Engineer currently at Entropik Technologies in Chennai, where I lead the backend development of real-time agentic AI moderation platforms. My work sits at the intersection of LLM orchestration, scalable Python backends, and production-grade AWS infrastructure.

Before Entropik, I built a full RAG platform at Infoware India — ingesting 10,000+ enterprise documents, integrating WhatsApp and voice-based LLM agents, and implementing memory layers for conversational continuity. I graduated from KIIT University, Bhubaneswar with a B.Tech in Computer Science.

Outside of work, I write about GenAI and backend systems on Hashnode, and I led Enactus KIIT as president — a 250-member entrepreneurship team that won India's National Cup and represented India at the World Cup in Kazakhstan. I'm a builder who writes, and a writer who builds.

🏆 Hack NITR 2024 — MongoDB Track Winner🏆 Hack4Bengal 2024 — Callchimp SDK Winner🌍 Enactus India National Cup Winner🌐 Enactus World Cup, Kazakhstan

QUICK STATS

CURRENT ROLEAssociate Backend Engineer @ Entropik Technologies
EDUCATIONKIIT University, B.Tech CSE · CGPA 7.5
LOCATIONChennai, India
CURRENTLY LEARNINGAdvanced LLM Evaluation · Distributed Systems
OPEN TOFull-time AI / Backend roles
AVAILABLE FOR OPPORTUNITIES

TECHNICAL PROFICIENCY

The Stack.

GENAI & LLM

LLMsRAGAgentic WorkflowsLangChainPrompt EngineeringEmbeddingsVector SearchQdrantDBpgvector

BACKEND & APIS

FastAPIExpress.jsREST APIsGraphQLWebSocketsAsync ProcessingCeleryMicroservices

CLOUD & DEVOPS

AWS EC2S3DynamoDBRDSCloudFrontECSDockernginxCI/CD

DATABASES

PostgreSQLMongoDBDynamoDBRedispgvectorQdrantDB

LANGUAGES

PythonJavaScriptNode.jsSQL
ML FUNDAMENTALS ·Scikit-learnNumPyPandasTensorFlowTransformer ArchitecturesETL PipelinesRBACJWTOAuth2pytestJest

WORK HISTORY

Experience.

Entropik Technologies Pvt. Ltd.

CURRENT
Backend Engineer — GenAI & Agentic Systems·Chennai, India·Nov 2025 – Present
  • ▸Led end-to-end backend development of a real-time agentic AI moderation platform using Python and FastAPI — API design, async processing, prompt orchestration, AWS deployment.
  • ▸Integrated LangChain for LLM orchestration and ElevenLabs for voice-based AI interactions.
  • ▸Designed and implemented an AI Moderator feature: prompt design, backend integration, structured LLM response handling for enterprise clients.
  • ▸Built cloud data pipelines using AWS S3, DynamoDB, and CloudFront.
  • ▸Implemented async background workflows with retries, idempotency, and graceful failure handling.

Infoware India

Backend Engineer — RAG Platform·Remote·Aug 2025 – Nov 2025
  • ▸Designed and delivered a full RAG platform for Kool.ai — 10,000+ enterprise documents ingested.
  • ▸Built scalable ETL and document ingestion pipelines using LangChain and QdrantDB.
  • ▸Integrated RAG pipeline with WhatsApp and a voice-based LLM call agent.
  • ▸Implemented short-term and long-term memory layers for conversational continuity.

SELECTED WORK

Projects.

🏆 Hack NITR · MongoDB Track

AI Meeting Assistant Backend

2024

REST API for meeting summarization — 500MB+ uploads via AWS S3 presigned URLs, Celery + Redis async processing, Whisper transcription + GPT-4 summarization, Docker + nginx + ECS auto-scaling.

PythonFastAPIPostgreSQLpgvectorCeleryAWSDocker

Farm Credit Scoring API

2024

Secure REST API with RBAC, JWT auth, Swagger/OpenAPI docs. Integrated SoilGrid and OpenWeatherMap APIs — reduced external API calls by 80%. Optimised MongoDB aggregations and geospatial queries.

Node.jsExpress.jsMongoDBRESTJWT

Healthcare Chatbot Backend

2024

Real-time chat API with WebSocket + GPT for AI-powered diagnosis suggestions. Appointment booking with calendar sync, SendGrid mailer, and schedule conflict resolution.

Node.jsWebSocketGPT APIPostgreSQL
🏆 Hack4Bengal · Callchimp SDK

Railway Track Health Monitoring API

2024

Flask REST API for track health prediction via vibration sensor data — 1M+ entries/day. Secured with JWT, Flask-Limiter; IoT ingestion with Pydantic validation and retry logic.

FlaskJWTIoTPydanticPython

WRITING

From My Desk.

Writing about GenAI, backend systems, and the craft of building.

Cover image for blog post: I Built a LogLens-CLI: That Uses No Vector DB, Here's How It Works
devtoolsPythonDevops

I Built a LogLens-CLI: That Uses No Vector DB, Here's How It Works

A few days back I was exploring why RAG is expensive and whether I could overcome it for structured data. I kept hitting the same wall, traditional RAG systems chunk text, embed it, and retrieve "simi

11 min3 May 2026
Writing

Building a Production-Grade Async OpenAI Status Tracker

1. Problem Statement & Goals The Problem OpenAI's status page (status.openai.com) publishes incident information via RSS feeds. There is no official webhook, push notification, or programmatic API to

5 min23 Feb 2026
Cover image for blog post: Understanding HTTP from First Principles

Understanding HTTP from First Principles

Every backend engineer uses HTTP. APIs, mobile apps, microservices, and they all "speak" it. Yet, most us learn HTTP as a list of status codes and methods to memorize. That approach misses the point. HTTP is not just a protocol; it is a set of carefu...

19 min5 Jan 2026
Cover image for blog post: Serialization and Deserialization: From First Principles

Serialization and Deserialization: From First Principles

Why Data Needs to Be Translated Before It Can Travel Modern backend systems are fundamentally about communication. A frontend written in JavaScript talks to a backend written in Python, Java, or Go.These systems may run on different machines, differe...

4 min4 Jan 2026
Cover image for blog post: Backend First Principles: The Art of Routing

Backend First Principles: The Art of Routing

Welcome back to the Backend First Principles series! Today, we’re stripping away the fancy frameworks (Express, FastAPI) to look at the skeletal structure of how the web actually works. At its core, every backend developer is a Digital Postmaster. Wh...

3 min3 Jan 2026
Cover image for blog post: My RAG Pipeline Keeps Failing: Here’s How I’m Fixing It
CRAGRAG llm

My RAG Pipeline Keeps Failing: Here’s How I’m Fixing It

I’ve been going deep on Retrieval-Augmented Generation (RAG) lately. The basic idea is simple and powerful: give a Large Language Model (LLM) access to your own data via vector search. It’s the foundation for almost every modern AI application that n...

7 min21 Aug 2025
View All Posts on Hashnode→

GET IN TOUCH

Let's Build
Something.

I'm open to full-time roles, freelance projects, and interesting conversations about AI. If you're building something ambitious in the GenAI space, I'd love to hear about it.

roulshivprasad@gmail.com