MADHUSUDAN RAI — DATA & AI ENGINEERINDEPENDENT THINKING. CONNECTED SYSTEMS.SELECTED PORTFOLIO / 2026

ENGINEERED WITH INTENT.
BUILT WITH CURIOSITY.

Turning complex data into
systems that think & do.

AIENGINEER.

An intricate chrome and cobalt sculpture, a visual study of connected intelligence — + CONNECTED INTELLIGENCE FIG. 01

RAG SYSTEMS
AGENTIC WORKFLOWS
DATA ENGINEERING

28° N / 77° E

Building the bridge between
what AI knows and what it can do.

Explore selected work

BASED IN NOIDA, INDIAIST · UTC +05:30

GENERATIVE AIHYBRID RETRIEVALAGENTIC SYSTEMSDATA & ANALYTICS

THE THREAD THROUGH MY WORK

Intelligence is a system.

InformationContextReasoningAction

01 / SELECTED WORK

Less artificial.
More useful.

(03)

A selection of systems I’ve built.
Made for the space between
information and action.

01 / KNOWLEDGE SYSTEMSPERSONAL PROJECT

Knowledge.
Connected.

A hybrid RAG engine that brings semantic understanding and keyword precision into the same conversation.

PythonPineconeDocling

HYBRID RAG SYSTEM / RETRIEVAL ENGINEERING

nexus / knowledge engine↗

YOUR KNOWLEDGE, CONNECTED Indexed

Find the signal.
Not just the keyword.

How does hybrid retrieval work?↵

Semantic

Keyword

▤

Context that understands you.

Relevant passages from connected documents.

⌘

Exact terms. Preserved.

Lexical signals, combined with meaning.

✳

One question.
A more grounded answer.

INTERFACE STUDY / PROJECT VISUALIZATION

SEMANTIC + LEXICAL ↗

02 / CONVERSATIONAL AI

orbit●POWERED BY LOCAL INTELLIGENCE

A conversation.
A connection.

I want to build something with AI.

✳

Let’s start with you.
What would you like to explore?

Automate a workflow ↗Understand my data ↗

Your next idea starts here↑

ILLUSTRATIVE INTERFACE

OLLAMA / LLAMA 3 / STREAMLIT

Conversations with purpose.

Local LLMs. Contextual responses. Structured lead capture.

03 / ENTERPRISE AUTOMATION

REVIEW AGENTAZURE / EVENT-DRIVEN

↗

TRIGGER RECEIVEDPull request opened

●

01async def review_pull_request(pr):

02 context = await fetch_diff(pr)

03 signals = await agent.analyze(

04 context, focus=["quality",

05 "security", "maintainability"])

06 await publish_feedback(signals)

DevOps MCP→AI agent→Inline feedback

ILLUSTRATIVE WORKFLOW / NOT PROPRIETARY CODE

MICROSOFT FOUNDRY / AZURE DEVOPS / MCP

From pull request to perspective.

Event-driven agents for code quality, security, and feedback.

THE THINKING CONTINUES IN THE CODE.Explore my GitHub

THE BUILD JOURNAL JUL — OCT 2026

Quietly building.
Thoughtfully connecting.

Experiments, working systems,
and the ideas between them.
A recent trail of public work.

ALSO IN THE NOTEBOOKHybrid RAG System 18 JUL 2026

Public repository activity from 1 July to 1 November 2026.
Reviewed 1 Nov · Related repositories grouped together.

The full notebook

02 / THE PLAYGROUND

Don’t just read it.
Get inside it.

A small window into how I build.
Follow the journey from a question
to a context-grounded answer.

n✳RETRIEVAL, EXPLAINED.CURATED DEMO

What would you
like to understand?

THE IDEA

Good retrieval is about finding the right context. Choose a question and follow the signals.

ARCHITECTURE DEMONSTRATIONNO LIVE MODEL OR API CALLS

THE PATH TO AN ANSWER

01 / UNDERSTAND

A query has both meaning and specific terms. Keeping both gives retrieval a richer starting point.

03 / THE HUMAN ELEMENT

MR / ENGINEER

mr.

MADHUSUDAN RAICURIOUS BY DEFAULT.

NOIDA, INDIAGet the résumé

Human curiosity.
Engineered possibility.

Hi, I’m Madhusudan. I build at the intersection of data, generative AI, and the things that make work work better.

I’m interested in the whole system: how documents become knowledge, how context becomes an answer, and how agents turn that answer into action.

My background spans analytics, data pipelines, hybrid RAG, and enterprise AI workflows on Azure. The common thread? Making complex information genuinely useful.

A CONNECTED THREAD

Evidence-grounded AI, structured knowledge, and automation with room for human judgment.

↗

THE FOUNDATIONB.Tech · Computer Science, AI & ML

Noida Institute of Engineering & Technology · AKTU

A FOUNDATION TO BUILD ON

Microsoft Certified:
Azure AI Fundamentals

I earned my Azure AI Fundamentals certification on 24 September 2026. It complements my hands-on work with retrieval systems, intelligent agents and Azure workflows.

AI workloadsMachine learningComputer visionNatural language processingGenerative AI

Credential details

Credential ID: 22CEA1C977B1EEFF
Certification number: 39A0CE-AA9C9A

THE TOOLKIT

Different tools.
Connected thinking.

Chosen for the problem.
Connected for the outcome.

Intelligence

RAG · LLM integration · CrewAI
Prompt engineering · NLP
Azure AI Foundry

REASON / GENERATE / ACT

Knowledge

Python · SQL · Pinecone
MongoDB · Docling · LlamaParse
ETL pipelines

PARSE / STRUCTURE / RETRIEVE

Connection

Microsoft Azure · DevOps · MCP
Logic Apps · Microsoft Graph
Git · Workflow automation

CONNECT / ORCHESTRATE / SHIP

Perspective

Power BI · Tableau · Pandas
NumPy · DAX · Power Query
KPI design

ANALYZE / EXPLAIN / INFORM

04 / EXPERIENCE

A little more
with every chapter.

From finding patterns in data
to building systems that reason.
Always learning. Always making.

SEP 2026 — NOWCURRENT

Brickred System

Associate Trainee

Building AI agents and enterprise workflows across the engineering lifecycle.

Event-driven code review with Microsoft Foundry, Azure DevOps MCP, and Logic Apps.
RAG pipelines that turn project notes and transcripts into traceable BRDs.
Multi-agent workflows with Microsoft Agent Framework, Graph APIs, and secure Azure execution.

JUN — AUG 2026

NorthCorp Software

AIML Trainee

Connected enterprise documents and data to retrieval and autonomous workflows.

Hybrid retrieval using dense embeddings and BM25 keyword search.
Document processing with LlamaParse and Docling, indexed in Pinecone.
CrewAI automation supported by MongoDB and SQL integration.

JUL — SEP 2025

Senpiper Technologies

Data Analyst Intern · Gurgaon

Made operational information easier to see, trust, and act on.

Interactive Power BI dashboards and KPI reporting.
Python and SQL workflows for cleaning, validation, and reporting automation.
Cross-functional analytics support and performance monitoring.

THE NEXT GOOD IDEA STARTS WITH A CONVERSATION.LET’S CONNECT / 05

Something
in mind?

DROP ME A LINEmadhusudan706rai@gmail.com

LinkedIn ↗+91 94737 41167 ↗GitHub Résumé Back to top

Let’s make something thoughtful. — Madhusudan

AIENGINEER.

Intelligence is a system.

Less artificial.
More useful.

Knowledge.
Connected.

Find the signal.
Not just the keyword.

Conversations with purpose.

From pull request to perspective.

Quietly building.
Thoughtfully connecting.

Don’t just read it.
Get inside it.

What would you
like to understand?

Understand the question

Connect the knowledge

Choose the useful context

Let evidence lead

Human curiosity.
Engineered possibility.

Microsoft Certified:
Azure AI Fundamentals

Different tools.
Connected thinking.

Intelligence

Knowledge

Connection

Perspective

A little more
with every chapter.

Brickred System

NorthCorp Software

Senpiper Technologies

Something
in mind?

AIENGINEER.

Intelligence is a system.

Less artificial. More useful.

Knowledge. Connected.

Find the signal. Not just the keyword.

Conversations with purpose.

From pull request to perspective.

Quietly building. Thoughtfully connecting.

Don’t just read it. Get inside it.

What would you like to understand?

Understand the question

Connect the knowledge

Choose the useful context

Let evidence lead

Human curiosity. Engineered possibility.

Microsoft Certified:Azure AI Fundamentals

Different tools. Connected thinking.

Intelligence

Knowledge

Connection

Perspective

A little more with every chapter.

Brickred System

NorthCorp Software

Senpiper Technologies

Something in mind?

Less artificial.
More useful.

Knowledge.
Connected.

Find the signal.
Not just the keyword.

Quietly building.
Thoughtfully connecting.

Don’t just read it.
Get inside it.

What would you
like to understand?

Human curiosity.
Engineered possibility.

Microsoft Certified:
Azure AI Fundamentals

Different tools.
Connected thinking.

A little more
with every chapter.

Something
in mind?