I'm an Agentic AI Developer focused on building autonomous, tool-using and multi-agent systems.
My current engineering interests sit at the intersection of:
I am an Agentic AI Developer building autonomous systems
< with MCP, A2A, LangGraph and Advanced Hybrid RAG >
:) I build systems that don't just generate text.
They reason → use tools → collaborate → evaluate → act.
LLMs
│
├── Agentic AI
│ ├── Multi-Agent Systems
│ ├── Agent-to-Agent (A2A)
│ └── Autonomous Workflows
│
├── Tool Connectivity
│ └── Model Context Protocol (MCP)
│
├── Retrieval
│ ├── Hybrid RAG
│ ├── Vector Search
│ └── Context Engineering
│
└── Production AI
├── Evaluation
├── FastAPI
├── GCP
└── Developer Tooling
My goal: Build AI systems that can move from understanding → reasoning → acting.
$ cat current_focus.txt
[+] Agentic AI
[+] Multi-Agent Architectures
[+] MCP
[+] A2A Protocol
[+] Advanced Hybrid RAG
[+] LLM Evaluation
[+] LangGraph
[+] LangChain
[+] Production AI Systems
┌───────────────────────────────────────────────────────────────┐
│ │
│ GITHUB PRs │
│ │ │
│ ▼ │
│ ┌─────────────┐ │
│ │ MCP SERVER │ │
│ └──────┬──────┘ │
│ │ │
│ ▼ │
│ ┌─────────────┐ │
│ │ LLM CLASSIFY│ │
│ └──────┬──────┘ │
│ │ │
│ ▼ │
│ RELEASE DOCUMENTATION │
│ │ │
│ ▼ │
│ ┌─────────────┐ │
│ │ CREATE PR │ │
│ └─────────────┘ │
│ │
└───────────────────────────────────────────────────────────────┘
An autonomous GitHub release documentation agent.
It fetches the previous 7 days of merged pull requests, classifies changes using an LLM, organizes release summaries and automatically opens a new GitHub pull request.
Stack
MCP · RAG · LangChain · FastAPI
USER REQUEST
│
▼
┌─────────────────┐
│ AGENT 01 │
│ CODE GENERATION │
└────────┬────────┘
│
▼
┌─────────────────┐
│ AGENT 02 │
│ AUTOMATED TESTS │
└────────┬────────┘
│
▼
┌─────────────────┐
│ AGENT 03 │
│ DEPLOYMENT │
└────────┬────────┘
│
▼
LIVE PRODUCT
A multi-agent software engineering pipeline built around the Agent-to-Agent (A2A) protocol.
Each agent specializes in a different stage of the development lifecycle.
Agent 01 → Generates frontend code Agent 02 → Executes automated tests Agent 03 → Handles deployment
Stack
A2A · LangGraph · FastAPI
KNOWLEDGE SOURCE
│
▼
┌─────────────────┐
│ QUESTION │
│ GENERATOR │
└────────┬────────┘
│
▼
RAG SYSTEM
│
▼
┌─────────────────┐
│ LLM-AS-A-JUDGE │
└────────┬────────┘
│
┌─────────┼─────────┐
▼ ▼ ▼
Faithful Correct Relevant
An automated evaluation platform for RAG systems.
The system generates 10 evaluation questions from supplied resources and evaluates RAG responses using an LLM-as-a-Judge approach.
Evaluation dimensions
Faithfulness · Correctness · Relevancy
Stack
LLM-as-a-Judge · RAG · FastAPI · Python
INBOX
│
▼
UNREAD EMAIL
│
▼
┌─────────────┐
│ RAG │
└──────┬──────┘
│
▼
KNOWLEDGE BASE
│
▼
LLM
│
▼
CONTEXT-AWARE REPLY
│
▼
GMAIL
A Gmail automation platform where users can create agents that retrieve relevant information from a knowledge base and automatically generate context-aware responses to unread emails.
Stack
MCP · LangChain · RAG · FastAPI
╭─────────────────────────────────────────────────────────────╮
│ │
│ 5000+ Developers & tech enthusiasts │
│ │
│ 20+ Web3 startups partnered │
│ │
│ 200+ Live developer bootcamps │
│ │
│ 500+ Developers mentored through GDSC │
│ │
│ 20+ Universities reached through workshops │
│ │
│ TOP 80 Worldwide ICP Hackathon │
│ │
│ 3rd PLACE ICP Hackathon — India │
│ │
│ SELECTED Google Developer Student Club Lead │
│ │
│ SELECTED Google Crowdsource Contributor │
│ │
│ SELECTED GenShot Lab Residency Program │
│ │
╰─────────────────────────────────────────────────────────────╯
Founded and scaled a Web3 & AI community of 5,000+ developers and technology enthusiasts, partnering with 20+ Web3 startups and leading 200+ live developer bootcamps.
Selected by Google to represent Google technologies through technical workshops and mentorship, reaching 500+ developers.
Built a Web3 e-ticketing platform that secured 3rd Prize in the ICP Hackathon across India and was selected among the top 80 projects from Asia for the Worldwide ICP Hackathon.
Delivered technical keynotes and blockchain development workshops across 20+ universities in India.
I don't want AI that only talks.
I want AI that can:
UNDERSTAND
↓
REASON
↓
USE TOOLS
↓
COLLABORATE
↓
EVALUATE
↓
ACT
The next generation of software won't just be written by developers.
It will be orchestrated by developers and intelligent agents.
pranav@ai-lab:~$ systemctl status pranav
● pranav.service - Autonomous Builder
Loaded: loaded
Active: active (running)
├── agentic-ai.service [RUNNING]
├── multi-agent.service [RUNNING]
├── mcp.service [RUNNING]
├── a2a.service [RUNNING]
├── rag.service [RUNNING]
├── experiments.service [RUNNING]
└── curiosity.service [RUNNING]
Status: Building...```