DATA SCIENTIST × SDE

ARNAV KAKKAR

I'm a |

I turn data into models, insights, and intelligent systems.

I build data-driven solutions using Python, SQL, machine learning, and AI. My work spans predictive modeling, anomaly detection, recommendation systems, and intelligent applications — from exploring data to building systems that can be put to use.

View My Work Resume
Arnav Kakkar - Data Science & AI Engineer
arnav@portfolio ~
$ python build_model.py
Loading dataset… ✓
Training model… ✓
Deployment… ✓
System ready_
01 // ABOUT

ABOUT ME

I’m an engineering student focused on Data Science, Machine Learning, and AI. I enjoy working with data, building models, and turning ideas into practical systems. My projects range from predictive modeling and analytics to RAG-based AI applications and real-time anomaly detection. I’m interested in taking projects beyond experimentation and building systems that are useful and applicable to real-world problems.

WHAT I BUILD
01

Data Science

Data analysis, feature engineering, and insights.

02

Machine Learning

Predictive models, classification, anomaly detection.

03

Generative AI

RAG, embeddings, vector search, AI assistants.

04

Data Analytics

Exploration, visualization and business insights.

02 // WORK

PROJECTS

Practical applications spanning predictive machine learning, generative retrieval, and real-time detection systems.

01 / PROJECT Healthcare AI

AIRA

AI Respiratory Intelligence & Risk Assessment

AI-powered respiratory risk assessment platform that combines real-time AQI monitoring, 24-hour AQI forecasting using GRU neural networks, anomaly detection, and personalized health recommendations to help users take preventive action before hazardous conditions occur.

  • 24-hour AQI forecast powered by a GRU neural network
  • Real-time environmental monitoring with anomaly alerts
  • Personalized respiratory preventive health recommendations
Python TensorFlow GRU Flask JavaScript Chart.js
View on GitHub ↗
aira — dashboard
AIRA Respiratory Intelligence Dashboard
CURRENT AQI
142 MODERATE
24H PREDICTED PEAK
178 UNHEALTHY
ANOMALY STATUS
NORMAL GRU OPT
24-Hour GRU Predictive Trajectory RMSE: 4.12
Now (T0) GRU Forecast (+24h)
02 / PROJECT Generative AI

RAG-Based AI Teaching Assistant

Generative AI & Semantic Vector Retrieval System

Built a retrieval-augmented generation system using LangChain and FAISS to answer queries from 500+ documents with improved accuracy and response quality.

  • Fast semantic vector search indexing over 500+ reference documents
  • High-relevance context retrieval with FAISS and LangChain pipelines
  • Accurate answer generation with hallucination mitigation and citations
Python LangChain FAISS OpenAI API NLP
View on GitHub ↗
rag-assistant — semantic-search
RAG-Based AI Teaching Assistant Interface
01 / DOCS
500+ Chunks
02 / EMBED
OpenAI API
03 / INDEX
FAISS Vector
04 / CHAIN
LangChain
Query: "Explain gradient clipping implementation in GRU models"
FAISS Similarity Score: 0.94 • 4 Context Chunks Retrieved
Synthesized from lecture documents & lab guides: Gradient clipping bounds gradients during backpropagation through time (BPTT), preventing exploding gradients in recurrent networks...
03 / PROJECT Machine Learning

Real-Time Anomaly Detection System

Streaming Time-Series & Outlier Identification

Developed a real-time anomaly detection pipeline using Isolation Forest with live monitoring, feature engineering, and a Streamlit dashboard.

  • Real-time anomaly detection pipeline using Isolation Forest algorithms
  • Live time-series streaming monitor with dynamic outlier boundary alerts
  • Interactive Streamlit dashboard for parameter tuning and metric inspection
Python Scikit-learn Streamlit Time Series
View on GitHub ↗
anomaly-detector — real-time-telemetry
Real-Time Anomaly Detection Dashboard
STREAM STATUS
LIVE 120ms
ISOLATION FOREST
ACTIVE n=100
DETECTED OUTLIERS
2 SPIKES ALERT
Live Telemetry • Isolation Forest Boundary Outlier Alert
Threshold 0.82 Outlier #1 Outlier #2

Credit Risk Prediction System

Machine learning model to classify applicants as good or bad credit risk using multiple algorithms and deployed with a Streamlit app.

Python Pandas Scikit-learn Streamlit
GitHub ↗

Fraud Detection System

Built a machine learning pipeline to detect fraudulent financial transactions using transaction patterns and deployed it as an interactive web app.

Python Pandas NumPy Scikit-learn Streamlit
GitHub ↗

Sales & Revenue Analysis

End-to-end business analysis uncovering revenue trends, customer insights, and discount impact to drive strategic decision-making.

Python Pandas Matplotlib Data Analysis
GitHub ↗

People Pages Recommendation System

Python-based recommendation system using similarity metrics and user interaction data to suggest relevant pages.

Python Pandas NumPy
GitHub ↗

Instagram Data Analysis

Exploratory data analysis on Instagram datasets to extract engagement patterns and behavioral insights.

Python Pandas Matplotlib Seaborn
GitHub ↗

Jarvis Voice Assistant

An interactive voice assistant utilizing speech recognition and automated task execution for system control.

Python Speech Recognition Automation
GitHub ↗

Secure Cloud Vault

Secure cloud storage and file management utility implementing authentication and encryption protocols.

Python Cloud Storage Cryptography
GitHub ↗

SignVox

Computer vision and gesture-to-speech assistive application bridging communication gaps through real-time sign detection.

Python OpenCV Computer Vision Deep Learning
GitHub ↗

Logic Looper

Algorithmic puzzle and logical reasoning project focusing on computational logic and structured problem solving.

Python Algorithms Logic
GitHub ↗
03 // TOOLKIT

Tools & Technologies

Categorized technologies utilized in practical data analysis, machine learning modeling, and full-stack AI deployment.

Python

Data • Core

Pandas

Data Manipulation

NumPy

Numerical Computing

SQL

Querying & Schemas

Jupyter Notebook

Interactive Research

Matplotlib

Visualization

Scikit-learn

Machine Learning

Time Series

Forecasting & Models

GRU / Neural Networks

Deep Learning

LangChain

LLM Pipelines

FAISS

Vector Search

OpenAI API

AI Integration

Streamlit

Interactive ML Apps

Flask

Web Backend & APIs

JavaScript

Frontend Logic
04 // GET IN TOUCH

LET'S BUILD
SOMETHING

Have a project, idea, or opportunity in Data Science, Machine Learning, or AI systems? Reach out directly or send a message.

Direct Email
arnavkakkar28.3@gmail.com
LinkedIn
arnavkakkar
GitHub
github.com/arnavkakkar
Location
Karnal, Haryana, India
Email Me Directly

Send a Message

Fill out the form below or connect via email.