Skip to content
View itsroshanharry's full-sized avatar
πŸ˜€
Building something cool
πŸ˜€
Building something cool

Block or report itsroshanharry

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
itsroshanharry/README.md

Hey πŸ‘‹ I'm Roshan Kumar

AI Engineer @ Accenture | Production AI Systems β€’ Agentic AI β€’ Distributed Systems

I build production AI systems at the intersection of software engineering, distributed systems, and modern AI architectures.

My work spans autonomous agents, multi-agent orchestration, Retrieval-Augmented Generation (RAG), evaluation frameworks, cloud infrastructure, scalable backend systems, and applied machine learning. I enjoy rapidly exploring new technologies, validating ideas through experimentation, and turning research into production-ready software.

Beyond shipping features, I care deeply about writing clean, maintainable code, designing resilient architectures, and building systems that are observable, reliable, and easy to evolve over time.

I see AI as an engineering acceleratorβ€”not something that replaces deep technical understanding. The real value comes from designing robust architectures, making sound engineering trade-offs, and using AI intentionally throughout the software development lifecycle.

I'm always looking to solve harder problems, build faster with better tools, and work with teams that value ownership, technical excellence, and continuous learning.

Building AI systems that don't just generate outputsβ€”but reason, collaborate, and deliver reliable outcomes.


πŸš€ What I'm Building

  • πŸ€– Autonomous Multi-Agent Systems
  • 🧠 Agentic AI Workflows
  • πŸ“š Production RAG Pipelines
  • πŸ”Œ Model Context Protocol (MCP)
  • 🀝 Agent-to-Agent (A2A) Communication
  • βš–οΈ LLM Evaluation & Guardrails
  • ☁️ Cloud-native AI on AWS, Azure & GCP
  • ⚑ Distributed Event-Driven Systems

πŸ› οΈ Tech Stack

AI Engineering

Programming

Python Node.js

Foundation Models

OpenAI Azure OpenAI AWS Bedrock

GPT Claude Llama Amazon Nova

Agentic AI Frameworks

LangGraph LangChain LlamaIndex CrewAI AutoGen Strands

AI Systems

RAG MCP A2A LLM Evaluation Guardrails

Machine Learning

scikit-learn PyTorch TensorFlow Feature Engineering Reinforcement Learning


Backend Engineering

Node.js TypeScript FastAPI Express REST API WebSockets


Cloud & Infrastructure

AWS

AWS Bedrock AWS Lambda Amazon API Gateway Amazon S3 Amazon DynamoDB

Azure

Azure OpenAI Azure Functions Azure Cosmos DB Gremlin API

Google Cloud

Google Cloud Cloud Spanner Vertex AI

Platform

Docker Serverless


Data & Messaging

PostgreSQL MongoDB Redis Apache Kafka


🧩 Engineering Interests

πŸ€– AI Systems

  • Agentic AI Systems
  • Multi-Agent Orchestration
  • Long-Running AI Workflows
  • Autonomous Agent Architectures
  • AI Infrastructure
  • LLM Evaluation & Benchmarking
  • AI Guardrails & Reliability
  • AI Observability
  • Semantic Caching

🧠 AI Architecture

  • Memory Architectures for Autonomous Agents
  • Retrieval-Augmented Generation (RAG)
  • Model Context Protocol (MCP)
  • Agent-to-Agent (A2A) Communication
  • Knowledge Graphs
  • AI Developer Tools

βš™οΈ Systems Engineering

  • Distributed Systems
  • Cloud Architecture
  • Backend Engineering
  • Event-Driven Systems

🀝 Let's Connect

πŸ’Ό LinkedIn
https://linkedin.com/in/itsroshanharry

🐦 X (Twitter)
https://twitter.com/itsroshanharry

πŸ“§ Email
roshankumarkola6@gmail.com


🌟 Featured Projects

πŸ€– Autonomous Swarm Agent

An autonomous multi-agent architecture where specialized AI agents collaborate to design, implement, test, review, and coordinate complex software tasks. Built using AWS Bedrock, Lambda, S3, API Gateway (WebSockets), and modern agentic AI patterns.

πŸ“š Modular RAG System

A production-oriented Retrieval-Augmented Generation framework with modular retrieval, evaluation pipelines, and extensible AI workflows.

πŸ”Œ AI Infrastructure & Agent Experiments

Exploring MCP, Agent-to-Agent (A2A) communication, memory architectures, semantic caching, evaluation frameworks, and long-running autonomous AI workflows.


πŸ’­ Philosophy

"The future isn't about building software that responds. It's about engineering systems that reason, collaborate, and continuously improve."

Pinned Loading

  1. Swarm-Agents Swarm-Agents Public

    An aws-production ready swarm of software development agents(architech, designer, developer, tester) with human-in-the-loop and real-time streaming implemented using strands.

    Python

  2. AR-Collections-agent AR-Collections-agent Public

    AI voice agent automating AR collections calls via LiveKit β€” extracting promise-to-pay commitments with risk-tiered dunning logic and grounding responses in real invoice data. Includes a FastAPI ev…

    Python 1

  3. saas-admin-eval saas-admin-eval Public

    Safety evaluation environment for AI agents on HUD's CUA framework. Claude 100% vs GPT-4o 63.3% resistance rate.

    Python

  4. Doppelganger-Openmetadata Doppelganger-Openmetadata Public

    Project DoppelgΓ€nger is an autonomous AI broker that solves the data privacy vs. utility trade-off. It detects PII in production and uses AI to generate "synthetic twins" with perfect statistical a…

    Python

  5. Multi-Agent-Research-System Multi-Agent-Research-System Public

    A multi-agent research system built using langraph.

    Python

  6. Automated-Lambda-Layer-Creation-Workflow Automated-Lambda-Layer-Creation-Workflow Public

    Automated lambda layers creation workflow that packages the dependencies in to a zip file in the required format of aws lambda layers.