Subak Kumar

Origins

The Shinobi's Path

AI & Data Science graduate from Saveetha University (SIMATS). I build production-grade intelligent systems — LLM-powered tools, RAG pipelines, ML classifiers, and full-stack analytics dashboards.

9 projects shipped. All open-source. All working. From code review assistants to interview coaches to churn prediction engines — each one solves a real problem.

Cisco & networking certified. Internship-tested at Orion EdTech. Presented at IIT Chennai and SIMATS Tech Star Summit. Currently open to work.

Projects Shipped

0.99

Best Model AUC

122K

Records Analyzed

∞

Lines Written

Disciplines

The Arsenal

⚡

AI & LLMs

Groq API LLM Integration Prompt Engineering RAG Pipelines ChromaDB Streaming APIs

🧠

ML & Data

XGBoost scikit-learn TensorFlow NumPy Pandas Matplotlib

⚙️

Backend

Python Flask SQLAlchemy REST APIs SQLite APScheduler

🌐

Networking

CCNA Cisco IOS Routing & Switching VLANs OSPF Netmiko

🛠️

Tools

Git & GitHub VS Code Jupyter Linux Wireshark Chart.js

📜

Languages

Python Java C/C++ SQL HTML/CSS JavaScript

Conquests

Boss Fights Won

Every project is a boss defeated — a real problem solved

AI · LLM

AI Interview Coach

Adaptive mock interview platform where AI adapts follow-up questions to your answers and generates a 5-dimension radar chart report.

Streaming AI + Voice Input + Radar Chart

Flask Groq Chart.js Web Speech API

View Repository

AI · Security

AI Code Reviewer

Reviews code for bugs, security vulnerabilities, and architecture issues. Returns categorized findings with line references and a quality score 0–100.

Detects SQL Injection · Hardcoded Secrets · Bad Arch

Flask Groq SQLite Prompt Eng.

View Repository

AI · RAG · NLP

RAG Knowledge Assistant

Chat with your own documents — PDFs, notes, markdown. Answers with exact source citations and relevance scores.

ChromaDB + Sentence Transformers + Cosine Similarity

ChromaDB Groq Flask PyPDF2

View Repository

ML · Classification

Customer Churn Prediction

Trains and compares 4 ML models to predict customer churn. XGBoost achieved AUC 0.9879 with 5-fold cross-validation.

AUC 0.9879 · XGBoost · 5-Fold CV

XGBoost scikit-learn Flask Matplotlib

View Repository

Full Stack · Analytics

Analytics Dashboard

Production-grade BI dashboard with live KPIs, revenue trends, category breakdown, regional analysis and sortable product tables.

7 API Endpoints · 8K Orders · Live Filters

Flask SQLite SQLAlchemy Chart.js

View Repository

MLOps · Monitoring

System Failure Monitor

Predicts system failures before they happen with actionable ETA alerts — "disk full in 14 minutes" — not just post-incident dashboards.

NumPy Trend Analysis · Live APScheduler Pipeline

Flask NumPy APScheduler Chart.js

View Repository

Chronicle

Battle Record

Data Analyst Intern

Orion EdTech · Tamil Nadu, India

Jan 2026 — Mar 2026

Performed revenue and operational data analysis using Python (Pandas), improving reporting accuracy by structuring and cleaning customer datasets
Created Matplotlib visualizations to identify sales trends and performance patterns, supporting data-driven business decisions
Extracted actionable insights and assisted in workflow optimization through structured performance analysis

Achievements & Recognition

Academic & Technical

Top Engineer — AI Workshop, IIT Chennai
Project Presentation — Brain Tumor Classification using EfficientNetB0, Tech Star Summit, SIMATS Engineering
Cisco Certified — Networking, Routing & Switching, Security

The Shinobi's Path

The Arsenal

Boss Fights Won

Battle Record

Send a Message