Skip to main content
Logo
Available for opportunities
Hi! I am Shubh Dwivedi

Building AI Systems That Solve Real Problems.

Data Science Student · ML Engineer · AI Researcher · building intelligent systems from research papers to production - with measurable, data-driven impact.

>
SD
model.accuracy
R² 0.91
regression score
dataset13.3K
samples · bengaluru
models.trained
5 experiments91%
CGPA
9.45
About

Engineer at the intersection of Data Science & AI.

I build intelligent, data-driven systems - from EDA and model training to production APIs and scalable architectures.

I'm Shubh Dwivedi - a Data Science student building production-grade ML systems and AI backends. My work spans the full pipeline: from raw data exploration and statistical analysis to model training, evaluation, and deployment.

I approach every problem as a researcher first. I've published work on AI governance (RSIF framework), built regression and classification models, engineered RAG pipelines with ChromaDB and vector embeddings, and shipped full-stack AI applications used by real users.

Currently exploring agentic AI systems, multi-agent architectures, and the intersection of machine learning with responsible AI governance.

Student → ResearcherML EngineerAI BuilderOpen to rolesKalyan, IN
Education
B.Sc in Data Science
B.K. Birla College, Kalyan · 2025–2028
9.45
CGPA
01
Research Papers
07+
Projects Built
03
Hackathons
Stack

The toolkit behind the systems.

A pragmatic, research-oriented stack - from raw data exploration and model training to intelligent APIs and production deployments.

core

Programming

01
PythonTypeScriptJavaScriptJavaC++R
core

Data Science

02
Exploratory Data AnalysisFeature EngineeringStatistical AnalysisData VisualizationHypothesis TestingExperimentation
core

Machine Learning

03
Scikit-LearnPandasNumPyMatplotlib & SeabornJoblibRegressionClassificationModel EvaluationCross-Validation
core

AI / LLMs

04
LangChainGroq SDKLlama 3Gemini APIRAG PipelinesPrompt EngineeringAgentic Workflows

Vector Databases

05
ChromaDBEmbeddingsKnowledge RetrievalSemantic Search

Backend

06
Node.jsExpressFastAPIFlaskREST APIsJWTWebSockets

Frontend

07
ReactNext.jsTailwind CSSFramer Motion

Databases

08
PostgreSQLSupabaseFirebase

Tools & Platforms

09
GitDockerJupyter NotebookVercelGoogle Maps APIPuppeteer
ML & Data Analytics

Data Science capabilities.

The methodology behind every data-driven project - from raw exploration to evaluated, deployed models.

5+
ML Models Compared
per experiment
3+
Datasets Worked With
structured & unstructured
6+
AI Projects Shipped
production-grade
1
Research Published
conference paper
Standard ML Pipeline
Data Collection
EDA
Feature Engineering
Model Training
Evaluation
Deployment

Exploratory Data Analysis

Distribution analysis, outlier detection, correlation heatmaps, and statistical profiling to understand data before modeling.

PandasMatplotlibSeaborn

Feature Engineering

Encoding, scaling, selection, dimensionality reduction, and domain-informed feature creation to maximize signal quality.

Scikit-LearnNumPyPandas

Model Training & Selection

Comparing regression, classification, and ensemble models with cross-validation. Selecting the best model for the task.

Scikit-LearnXGBoostCross-Validation

Evaluation & Metrics

Rigorous evaluation using R², RMSE, MAE, precision, recall, F1-score, and confusion matrices - not just raw accuracy.

Model EvaluationJoblibStatsmodels

AI & LLM Integration

Building RAG pipelines, agentic workflows, and LLM-powered systems on top of structured and unstructured data.

LangChainChromaDBGroqLlama 3

Production Deployment

Serializing trained models with Joblib, exposing inference via FastAPI endpoints, and building interactive dashboards.

FastAPIJoblibCustomTkinterDocker
Projects

Selected work.

From ML systems and data pipelines to full-stack AI products - every project treated like a real product.

Featured Projects

ML Regression · Data Science Flagship
Bengaluru House Price Prediction
01
0.91
R² Score
₹7.8L
RMSE
₹5.2L
MAE
13.3K
Training Rows

End-to-end machine learning system that predicts Bengaluru residential property prices. Covers the full data science pipeline - EDA, feature engineering, model training, evaluation, and an interactive desktop dashboard for real-time inference.

  • Exploratory Data Analysis with outlier detection (3σ rule)
  • Feature Engineering: price_per_sqft, location grouping, encoding
  • 5-model comparison: Linear, Ridge, Lasso, Random Forest, XGBoost
  • Best model: Random Forest - R² = 0.91, RMSE = ₹7.8L
  • Cross-validation (k=5), MAE, RMSE, R² evaluation suite
  • Interactive desktop dashboard via CustomTkinter for live prediction
  • Model serialization with Joblib for production deployment
PythonPandasNumPyScikit-LearnMatplotlibSeabornJoblibCustomTkinterJupyter Notebook
Trade Intelligence · RAG · Multi-Agent
GlobeX AI
02
RAG
Architecture
ChromaDB
Vector DB
NestJS
LLM Backend
Hackverse 2026
Hackathon

AI-powered global trade analytics platform built during Hackverse 2026. Processes real export/import datasets, embeds trade data with ChromaDB, and surfaces market insights via an LLM-powered copilot with RAG architecture.

  • RAG pipeline: ChromaDB embeddings over real trade datasets
  • LLM-powered trade copilot (NestJS + TypeScript backend)
  • Export/import trend analysis and market intelligence
  • Structured data ingestion with embedding generation
  • Interactive dashboard for trade visualization (React frontend)
  • AI-driven insights on tariffs, trade corridors, and risk signals
TypeScriptNestJSChromaDBReactLangChainGroq SDKPostgreSQLDocker
AI Safety Platform · Real-Time Systems
Sahasya
03
Hackathon
Google Solution Challenge 2026

Full-stack women safety platform with instant emergency response, live GPS tracking, and fail-safe Telegram alerts. Designed for real-world use with a scalable backend and offline-capable features.

  • One-touch SOS activation with real-time alert dispatch
  • Live GPS tracking & emergency contact integration
  • Safety dashboard with incident history & streaks
  • Offline-ready women legal rights companion
ReactNode.jsTypeScriptSupabaseTelegram APIGoogle Maps APIJWT Auth
Privacy AI · Zero-Retention Architecture
ZeroReuse AI
04
Hackathon
TechXpression × CSI · Round 1

Privacy-focused AI image processing platform with zero-retention architecture. Uploaded images are never stored, reused, or used for training — processed entirely in isolated in-memory pipelines.

  • Stateless temporary AI processing - no persistent storage
  • Auto-deletion execution pipeline after each request
  • Isolated in-memory processing for data privacy
  • Zero-retention guarantee enforced at architecture level
ReactNode.jsFastAPITailwind CSSAI ModelsSupabaseFlask
Accessibility Engine · AI Analysis
AccessNX
05
Hackathon
TechXpression × CSI · Round 2

AI-powered accessibility analysis engine that scans websites, detects WCAG 2.1 issues, and generates targeted remediation suggestions using Llama 3 via Groq.

  • Dynamic DOM accessibility scanning via Puppeteer + Axe-Core
  • AI-generated accessibility fixes using Llama 3
  • WCAG 2.1 compliant issue detection & categorization
  • Context-aware remediation workflows per element type
Node.jsExpress.jsPuppeteerAxe-CoreGroq SDKLlama 3JavaScript

Experiments & Side Projects

Taskify
AI Task Manager

Voice-to-task AI platform using natural language parsing with Groq + Llama3.

ReactFastAPIPythonSupabase
Qubit AI
Interactive Quiz Platform

KBC-inspired AI quiz platform with real-time scoring, dynamic timers, and leaderboards.

ReactNode.jsExpress.jsFramer Motion
Vecna Curse Check
Cinematic AI Experience

Stranger Things-inspired interactive horror experience using camera input, AI analysis, and Firebase leaderboards.

ReactTypeScriptFirebaseGemini API
Research

Published work.

Bridging engineering and academic thought on responsible AI systems.

Conference Paper · 2026 · B.K Birla College, Kalyan

Reconceptualizing Accountability in AI-Driven Startup Ecosystems

A framework - RSIF (Responsible Sustainable Innovation Framework) - that redefines how early-stage AI companies should embed accountability across product, governance and lifecycle decisions.

International Student Research Conference
RSIF framework
AI governance
Hackathons

Shipped under pressure.

Where I sharpen architecture instincts and team execution.

2026

Google Solution Challenge 2026

Team Lead · Backend Architect
  • Designed scalable backend & data flow
  • Led architecture planning
  • Cross-functional team coordination
2026

TechXpression × CSI Hackathon 2026

Team Lead · Backend Engineer
  • PPT round qualified
  • Offline 25 hour hackathon participation
  • Built core API + auth system
2026

HackVerse 2026

Team Lead · Backend Engineer · AI Agent Developer
  • Led architecture and development of GlobeX AI
  • Built autonomous AI workflows and backend infrastructure
  • Designed retrieval pipelines for market analysis and insights
Credentials

Certifications & training.

IBM SkillsBuild - AI Fundamentals

IBM

IIT Bombay Techfest - AI Workshop

IIT Bombay

Microsoft AI Skills Passport

Microsoft

HCL GUVI - AI Summit

HCL × GUVI

Data Analysis & Visualization

B.K. Birla College, Kalyan
Contact

Let's build something unforgettable.

Open to internships, research collaborations, and ambitious product work.