← Back to Profile
Step 01

Hi, I'm Ramesh Das.

An Agentic AI Engineer and Backend Architect with 6+ years building production-grade AI platforms, scalable APIs, and distributed backend systems — currently leading AI development at KLIKY AI.

6+
Years Experience
9+
Major Projects
3
Companies
7+
Open Source Repos
● Open to opportunities Agentic AI Engineer LLM Engineer Backend Architect AI Platform Engineer Generative AI Developer Nepal · Remote

Use the sidebar to walk through this presentation step by step, or scroll freely. Each section tells you something specific about how I work and what I build.

Step 02

What I Specialize In

My core focus is at the intersection of Agentic AI systems and production backend engineering. I don't just prototype — I build systems that run in production at scale.

Primary Focus

Agentic AI & LLM Engineering

Designing and deploying autonomous AI systems that reason, plan, and act. Multi-agent workflows, tool-calling LLMs, RAG pipelines, and real-time AI orchestration layers on top of production infrastructure.

LangChain LlamaIndex Multi-Agent RAG Pipelines Tool-Calling LLM Orchestration Vector Search GPU Inference
Power Anchor

Backend Architecture at Scale

High-performance async APIs, microservices architecture, event-driven systems, and cloud-native deployments. The engineering foundation that makes AI systems actually work in production.

FastAPI Django Microservices Kafka Celery Async APIs WebSockets RBAC
Supporting Layer

Generative AI Integration

Integrating diffusion models, speech AI, video/audio processing, OCR, and multi-model routing into production platforms. Experienced with GPU inference services and model deployment pipelines.

ComfyUI Replicate Hugging Face OpenAI APIs RunPod FAL.ai Ollama DeepSeek
Full Stack Delivery

End-to-End Platform Building

From React/Next.js frontends to DevOps pipelines — I deliver complete products. SaaS billing, payment integrations, admin dashboards, analytics systems, and cloud deployment.

Next.js React Docker Kubernetes Stripe Firebase AWS GCP
Step 03

Full Tech Arsenal

Every tool here has been used in production — not just learned for a tutorial.

Agentic AI & Generative AI
LangChainLlamaIndex OpenAI GPT APIsHugging Face OllamaDeepSeek ComfyUIReplicate DifyLangflow RAGFlowBrowser-Use RunPodFAL.ai Multi-Agent SystemsTool-Calling LLMs RAG PipelinesVector Embeddings Prompt EngineeringGPU Inference Diffusion ModelsSpeech-to-Text/TTS OCRText-to-Image Face SwapVideo/Audio AI
Backend Development
PythonFastAPI DjangoNode.js Express.jsCelery Apache KafkaDebezium CDC GraphQLREST APIs WebSocketsAsync MicroservicesJWT/OAuth 2.0 RBACTypeScript
Databases & Vector Stores
PostgreSQLMongoDB RedisMySQL SupabasePinecone WeaviateVector Search Semantic SearchEmbeddings
Cloud, DevOps & Infrastructure
DockerKubernetes AWS EC2/S3/LambdaGCP GitHub ActionsNginx CI/CDSentry VPSGPU Servers
Auth, Frontend & Integrations
React.jsNext.js TypeScriptTailwind CSS Auth0Firebase StripePayPal TwilioDeepL WebhooksDjango Allauth
Step 04

How I Build Systems

I design for production from day one. Every system I build follows microservices architecture, async-first design, and cloud-native deployment — so it can scale without a rewrite.

Architecture Principle 1

Microservices & Event-Driven

Each service owns its domain and communicates via APIs or Kafka event streams. AI services (image generation, speech, OCR) scale independently from user management or payment services. Fault in one doesn't cascade.

Architecture Principle 2

Agentic AI Layer on Top

LLM orchestration, tool-calling, and multi-agent workflows sit as an orchestration layer above the microservices backend. Agents call tools (APIs, databases, external services) via structured interfaces.

Architecture Principle 3

Async-First, GPU-Aware

Heavy AI tasks (inference, image gen, video processing) are queued via Celery to GPU workers on RunPod or AWS Lambda. Clients get real-time updates via WebSocket streams — never blocking the main API.

Architecture Principle 4

Security & Observability

JWT/OAuth 2.0, RBAC, CSRF protection, SSL, and Auth0 across every service. Sentry for error monitoring, structured logging, and Kubernetes health checks for production reliability.

WinstaAI System Architecture

Web / Mobile Client
API Gateway (FastAPI)
Auth Service (JWT/Auth0)
Microservices
AI Orchestration
LangChain · Agents
Image / Video AI
ComfyUI · Replicate
Speech / OCR AI
TTS · ASR · OCR
User Service
Django · RBAC
Payment Service
Stripe · PayPal
Admin Dashboard
Analytics · Credits
PostgreSQL
Redis Cache
Pinecone / Weaviate
MongoDB
Docker · Kubernetes
AWS / GCP
Kafka Event Stream
Celery · GPU Workers
Sentry · CI/CD
Step 05

Work Experience

6+ years across AI SaaS platforms, multilingual AI tools, healthcare systems, and eCommerce — always shipping to production.

Lead AI Full-Stack Developer
KLIKY AI
Jan 2024 – Present

WinstaAI — Agentic AI SaaS Platform

Production AI platform with Generative AI, agentic workflows, and multi-model LLM orchestration.

  • Architected backend with FastAPI, Django, microservices, and async task workflows
  • Built agentic AI pipelines — tool-calling, autonomous execution, decision-based orchestration
  • Integrated text-to-image, face swap, OCR, video/audio AI, and multi-model LLM routing
  • Built intelligent admin: AI service management, credit systems, usage analytics, billing
  • Integrated Stripe, PayPal, RunPod, FAL.ai, CI/CD pipelines
Full-Stack Developer
AJATH INFO TECH
Jun 2021 – Dec 2023

Dibbly — Multilingual AI Video Translation

AI-powered speech-to-text, TTS, real-time transcription, multilingual dubbing. DeepL translation, Firebase auth, REST APIs, secure session handling.


Paperport — AI Presentation Generator

LLM-powered tool generating structured presentations and roadmaps from unstructured text and images. Firebase auth, OTP, push notifications, REST APIs.

Full-Stack Developer
SARAJ SYSTEM
Dec 2019 – May 2021

Khantailor — Custom Tailoring eCommerce

eCommerce platform with measurements, Stripe/PayPal, Docker/Nginx, JWT/CSRF security.


PathHub — HIPAA Medical Platform

HIPAA-focused diagnostic reporting platform serving a large hospital network — real-time imaging, secure access, scalable modules.


OZMED — Online Learning Platform

Subscription learning with mock exams, progress tracking, and secure content delivery.

Step 06

WinstaAI — Deep Dive

My current flagship project. A production agentic AI SaaS platform built entirely from scratch — every service designed, architected, and shipped by me.

Production
Live at KLIKY AI
12+
AI Services Integrated
6
Microservices

AI Services Built

Text-to-Image
ComfyUI + Replicate + Stable Diffusion on GPU
Face Swap
Privacy-focused processing microservice
Speech-to-Text
Real-time ASR with streaming response
Text-to-Speech
Multi-voice TTS with async delivery
OCR Pipeline
Document extraction and structuring
Video / Audio AI
Processing workflows via RunPod GPU
LLM Orchestration
Multi-model routing: GPT, DeepSeek, Ollama
Agentic Workflows
Tool-calling agents with LangChain
Admin Dashboard
Service mgmt, credits, analytics, billing

Backend & Infrastructure Stack

Backend APIs
FastAPIDjango CeleryWebSockets REST APIsAsync
AI / Inference
LangChainLlamaIndex ComfyUIRunPod FAL.aiOpenAI
Data & Storage
PostgreSQLRedis MongoDBPinecone Weaviate
Infrastructure & Payments
DockerKubernetes AWSGitHub Actions StripePayPal
Step 07

Skills Breakdown

An honest, data-driven view of where I spend my engineering time and depth.

Agentic AI Platforms
100%
FastAPI / Python Backend
100%
RAG Pipeline Architecture
92%
LLM Orchestration
90%
Generative AI Integration
88%
Microservices / Kafka
82%
Cloud / GPU Inference
76%
Full-Stack Frontend
68%
Technology Category Distribution

Skill Growth Over Career (2019 → 2025)
Step 08

Open Source Contributions

Production-ready templates that teams can use to skip boilerplate and go straight to building. All maintained under fullstack-open-source.

FastAPI Backend

Python · PostgreSQL · Docker · K8s

Enterprise-grade API with JWT Auth, RBAC, async workers, and Kubernetes support.

FastAPI PostgreSQL Docker
View on GitHub →

Node.js Backend

Express.js · Prisma · PostgreSQL · Docker

Production Express.js backend with JWT Auth, Prisma ORM, and Kubernetes support.

Node.js Prisma Docker
View on GitHub →

Django Backend

DRF · PostgreSQL · Auth · Admin

Full-featured Django REST backend with auth, admin panel, and production settings.

Django DRF PostgreSQL
View on GitHub →

Next.js Frontend

App Router · TypeScript · Tailwind CSS

Modern Next.js starter with App Router, TypeScript, Tailwind, and API integration patterns.

Next.js TypeScript Tailwind
View on GitHub →

React.js Frontend

Vite · TypeScript · React Router

React starter with Vite, TypeScript, React Router, and state management patterns.

React Vite TypeScript
View on GitHub →

More Coming Soon

Active Development

More AI and backend templates in progress — LangChain starter, FastAPI + RAG boilerplate, and more.

LangChain RAG Agentic AI
Watch the org →
Step 09

Let's Work Together

I'm open to Agentic AI Engineer, LLM Engineer, Backend Architect, and Full-Stack AI Developer roles — remote, contract, or full-time. Response within 24 hours.

Available for: Agentic AI platforms, LLM integrations, RAG pipeline architecture, FastAPI/Django backend systems, full-stack AI SaaS, microservices design, and GPU inference pipelines. Remote. Contract or full-time.