Aarica Raj AaricaMatrix

About Me

class AaricaRaj:
    def __init__(self):
        self.role       = "BCA Student — Data Science & AI (learning)"
        self.education  = "BCA; Final Year, Netaji Subhas University"
        self.location   = "Jamshedpur, India"
        self.contact    = "shrivastavarchana57@gmail.com"
        self.domain     = ["Data Analytics", "NLP", "RAG Systems",
                           "Deep Learning", "MLOps basics", "Multimodal AI"]
        self.stack = {
            "languages"      : ["Python", "SQL", "MySQL",
                                "JavaScript (basic)"],
            "data_analytics" : ["Pandas", "NumPy", "Matplotlib",
                                "Seaborn", "Excel", "Power BI basics"],
            "ai_ml"          : ["LLMs", "RAG", "Transformers",
                                "Computer Vision", "Generative AI"],
            "frameworks"     : ["LangChain", "HuggingFace", "PyTorch",
                                "Scikit-learn", "Flask", "Gradio"],
            "vector_search"  : ["FAISS", "all-MiniLM-L6-v2",
                                "Semantic Embeddings"],
            "cloud_apis"     : ["Azure AI", "HuggingFace Spaces",
                                "Groq API", "Gemini API", "REST APIs"],
            "mlops"          : ["Docker", "Git", "GitHub",
                                "Jupyter", "Google Colab"],
        }

    def current_focus(self):
        return [
            "Data analysis with Python, SQL and visualization tools",
            "Prepping for Data Analyst roles; EDA, dashboards, storytelling",
            "Personal RAG pipeline projects; FAISS + LangChain + Groq",
            "Fine-tuning DistilBERT and RoBERTa for NLP tasks",
            "Learning the ML lifecycle hands-on: train, containerize, deploy",
        ]

    def open_to(self):
        return [
            "Data analytics projects and collaborations",
            "AI/ML learning, research, and open-source contributions",
            "Transformer fine-tuning and RAG systems",
            "Student researcher and internship opportunities",
        ]

Technical Stack

Data Analytics & Visualization

AI / ML & Deep Learning

NLP, LLMs & Generative AI

Web, Deployment & MLOps

Cloud & APIs

Tech Stack Orbit

Featured Projects

🌿 EcoTrack AI

AI-powered carbon footprint tracker; hackathon project

FastAPI Scikit-learn Groq AI Python Anomaly Detection

Predicts carbon footprint using a scikit-learn model trained on synthetic data I generated
Groq AI integration for a simple chatbot-style eco coach
Basic anomaly detection to flag unusual emission patterns
Built end-to-end for a hackathon; solo project

🤖 RAG PDF Chatbot

Ask questions over any PDF using semantic search + LLM

LangChain FAISS HuggingFace Embeddings Python Gradio

Built to learn the RAG pattern: PDF ingestion, FAISS vector store, semantic retrieval, LLM answer generation
Uses HuggingFace embeddings for document understanding
Simple Gradio UI; ask questions over your own documents
Deployed on HuggingFace Spaces for others to try

🎭 Sentiment Analysis with BERT

Transformer fine-tuning practice: emotion and sentiment classification

DistilBERT DistilRoBERTa HuggingFace Python

Fine-tuned DistilRoBERTa for 7-class emotion classification; 91.3% accuracy on a held-out test set
Fine-tuned DistilBERT separately for binary sentiment
Built to practice the HuggingFace fine-tuning workflow end to end
Deployed as a live HuggingFace Space demo

🌸 Iris MLOps Pipeline

Learning end-to-end ML deployment, with a 3D particle UI

Python Flask Docker Random Forest REST API

Random Forest classifier wrapped in a Flask REST API and containerized with Docker
A practice run through the ML lifecycle: data prep, training, API deployment
Ships with a 3D particle-effects UI; not your average Iris project!

🕹️ Quantum Neural AR

3D neural network you control with your hands; no install needed

Three.js MediaPipe Hands Web Audio API WebGL HTML

Real-time 3D neural network visualization controlled entirely by hand gestures
Uses MediaPipe for hand tracking and Web Audio API for sound interaction
Runs 100% in the browser; no backend, no install
Also responds to claps and snaps via microphone input

🎮 Suzume Game

A browser game built with pure HTML, CSS and JS

JavaScript CSS HTML Creative Coding

A fully playable game inspired by the Suzume universe
Built with vanilla JS and CSS animations; no frameworks
Focuses on smooth interactions and a creative browser-based experience

🌀 TubesCursor

Neon tube cursor animation using WebGL

Three.js WebGL JavaScript Canvas

A neon tube effect that follows your mouse in real time
Built with Three.js and WebGL for smooth GPU-accelerated rendering

🐢 Turtle Animations

Math-inspired generative art with Python's Turtle library

Python Turtle Graphics Generative Art Math

A collection of geometric patterns and animations
Covers physics simulations, math-art and procedural generation
Good starting point if you want to explore creative coding with Python

🖼️ AI Image Captioning with BLIP

Multimodal AI exploration; Vision Transformer + BERT

BLIP Vision Transformer BERT PyTorch HuggingFace

Used Salesforce's pretrained BLIP model (247M params, ViT encoder + BERT decoder) to learn how vision-language captioning works
Supports both prompt-guided and unconditional captioning
Built to explore cross-modal attention between visual and text representations

More on GitHub!

Project Timeline

RAG Pipeline Architecture

Project Highlights

Metric	Value	What it means
Deployed Demos	4+	Personal projects live on HuggingFace Spaces
Best Model Accuracy	91.3%	DistilRoBERTa fine-tuned for 7-class emotion classification, on my own held-out test split
RAG Response Time	~0.5s	With Groq API in my RAG PDF chatbot (single-user demo, not load-tested)
Model Scale	247M params	Size of the pretrained Salesforce BLIP model I used
Certification	Azure AI	Microsoft & LinkedIn Learning, 2025
Vector Search	FAISS	Semantic retrieval on custom document corpora

Skill Proficiency

Education & Certifications

Degree	Institution	Status
Bachelor of Computer Applications (BCA)	Netaji Subhas University, Jamshedpur	Final Year, 5th Semester (Expected July 2027)

Certification	Issuer
Azure AI Fundamentals	Microsoft

GitHub Analytics

Let's Build Together

🔗 Professional

💻 Development

📧 Contact

📧 Email | 📍 Jamshedpur, India 🇮🇳

🔭 Currently building: Data analytics skills; SQL, EDA, dashboards, storytelling with data. Also building personal RAG projects and fine-tuning transformers on the side!

💡 Philosophy: Understand the data first; then let the model do the talking.

⭐ If my projects helped you, a star means a lot!

Navigation Menu