Available for Full-Time & High-Impact AI Roles

Hi, I'm Rahul Kumar Pradhan

AI & Machine Learning Engineer|

I build end-to-end AI systems that bridge the gap between experimental research and scalable production solutions. Focused on turning complex data into measurable business ROI through rigorous engineering, reproducible MLOps, and model explainability.

Explore Projects
Rahul Kumar Pradhan
Rahul Kumar Pradhan @rkpcode • AI / ML Engineer
Regulatory XAI
SHAP Auditability
Production ML
95%+ Model Recall
99.91%
ROC-AUC Score
Customer churn discrimination (98.42% Recall)
95%+
Credit Recall
Minimized exposure in automated risk scoring
90.54%
NLP Accuracy
IMDB review sentiment trained in ~7s
85.2%
Anomaly Precision
Margin leak detection engine in retail data
18+
Repositories
Production, research & open-source ML codebases
Recruiter Fast-Track

Why Hire Rahul for Your Team?

I don't just fit models on CSVs — I engineer resilient, explainable, production-ready systems tailored to business ROI.

Production-First Mindset

Moving beyond Jupyter notebooks to build reproducible pipelines using DVC for data versioning, MLflow for experiment tracking, and Docker + FastAPI for containerized low-latency serving.

DVC MLflow Docker FastAPI CI/CD

Explainable AI (XAI)

AI should never be a black box. I leverage SHAP values and interpretability trees to guarantee regulatory compliance, build stakeholder trust, and provide transparent decision justifications in high-stakes FinTech.

SHAP Values Feature Importance Compliance Model Transparency

Interdisciplinary Edge

My background in Civil Engineering instills mathematical rigor and structured systems thinking, while a Master's in Mass Media allows me to decode human behavior, churn patterns, and NLP tasks intuitively.

Systems Thinking Behavioral Analytics Mathematical Rigor
Featured Engineering

Production Systems & Research

Explore real-world end-to-end applications, deep learning architectures, and data intelligence tools.

🎯
Customer Retention ML

Customer Churn Prediction

99.91% ROC-AUC 98.42% Recall • 92.57% Precision with LightGBM
Production Pipeline LightGBM • SHAP • Streamlit

Explainable Churn Risk Scoring Engine

Engineered an end-to-end churn prediction architecture across 7 comparative models. Applied business-driven threshold optimization (0.247 threshold) targeting top 20% highest risk users with SHAP waterfall explainability.

  • Comprehensive 7-model tournament (LightGBM, XGBoost, CatBoost, Random Forest)
  • Integrated SHAP waterfall & summary plots for single/batch risk factors
  • Interactive Streamlit dashboard supporting live single & bulk CSV scoring
LightGBM XGBoost SHAP Streamlit Scikit-Learn Kaggle API
GitHub Repo
📈
Quantitative Deep Learning

NIFTY 50 Forecaster

Seq2Seq LSTM 7-Day recursive multi-step price forecast

Deep Learning Financial Trend Forecaster

State-of-the-art hybrid Deep Learning application predicting NIFTY 50 index movements 7 days ahead using Encoder-Decoder Seq2Seq LSTM architectures and live yfinance streaming.

  • Seq2Seq LSTM & Multivariate technical indicator processing
  • Online learning capabilities to retrain on live market volatility
TensorFlow LSTM Seq2Seq yfinance Streamlit
Code
🎬
Production NLP Pipeline

NLP Review Analyzer

90.54% Accuracy 50,000 reviews trained in ~7 seconds

IMDB Sentiment Classification Pipeline

Production-ready modular NLP pipeline with TF-IDF vectorization, specialized negation handling (e.g. "not good" classification), and a deployed live Streamlit application.

  • Modular component architecture (ingestion, transform, trainer, monitor)
  • Live interactive web demo deployed on Streamlit Cloud
NLP TF-IDF Scikit-Learn Streamlit Python
Live Demo Repo
🎬
Multi-Agent GenAI

AutoReel_AI — Video Creator

100% Autonomous Single Topic Input → 1080×1920 Published Reel
Multi-Agent Pipeline CrewAI • Llama-3.3 • Flux.1

Autonomous Hinglish Short-Form Reel Generator

End-to-end multi-agent orchestration taking a single topic and generating research (DeepSeek), Hinglish scripts (Llama-3.3-70B), voiceovers (ElevenLabs), visuals (Flux.1), and an auto-synced rendered vertical MP4 reel with MoviePy.

  • CrewAI multi-agent workflow orchestrating two LLMs and 3 generative APIs
  • Docker containerized REST API with background task queues
  • Automated 1080×1920 video composition with beat and voice alignment
CrewAI DeepSeek Llama-3.3-70B ElevenLabs Flux.1 MoviePy Docker
GitHub Repo
⚡
Distributed Pipeline

Video Scraper Engine

Concurrent Pipeline Harvester & Worker with Supabase recovery

Autonomous Video Ingestion & CDN Sync

Production-ready video ingestion system with intelligent multi-page discovery, crash-resilient PostgreSQL state tracking in Supabase, and Bunny Stream CDN syncing.

  • yt-dlp buffer limits & User-Agent rotation for WAF evasion
  • Real-time Gradio telemetry dashboard with live worker stats
Python Gradio Supabase PostgreSQL yt-dlp Bunny Stream
Code
🛡️
FinTech Risk Engine

Loan Default Guard

95%+ Recall Minimized Credit Default Risk & Auditable Reasons
Production Ready Python • XGBoost

Auditable Loan Default Prediction System

Solved the critical dilemma of credit risk and regulatory non-compliance in automated loan underwriting. Achieved 95%+ Recall on high-risk borrowers while generating SHAP-powered explainability reports for transparent loan rejection audits.

  • Engineered XGBoost classification with hyperparameter optimization
  • Integrated SHAP for individualized feature-level rejection explanations
  • Deployed an interactive Streamlit risk assessment console
XGBoost SHAP (XAI) Streamlit Scikit-Learn Pandas
Case Study ↗ Repo
📊
Sales Intelligence

Retail Margin Intel

85.2% Precision 15+ hrs/week manual audits saved

Automated Margin Leak Anomaly Engine

Manual audits for margin erosion were extremely costly and slow. Built an anomaly detection pipeline that flags margin leaks with 85.2% Precision.

  • Handled massive class imbalance using SMOTE
  • Containerized low-latency scoring API using FastAPI
XGBoost SMOTE FastAPI Docker
Case Study ↗ Code
🏗️
Computer Vision

Industrial Steel QA

Pixel-Level U-Net Automated surface defect segmentation

PyTorch Defect Segmentation Pipeline

Replaced slow and inconsistent manual surface inspection in heavy manufacturing with a custom U-Net deep learning segmentation architecture.

  • Trained in PyTorch with image augmentation & dice loss
  • Packaged with Docker for edge manufacturing deployment
PyTorch Computer Vision U-Net Docker
Case Study ↗ Code
⚡
Live Cloudflare Fleet

Client Web Factory

100+ Live Sites Multi-client automated generation & edge deployment

Cloudflare Pages High-Speed Factory

End-to-end Python automation pipeline that ingests client branding, extracts color palettes, generates high-converting responsive websites, enforces strict 14-point pre-deploy verification, and deploys to Cloudflare Pages edge network.

  • Automated color extraction, font pairing & design token injection
  • Instant edge deployment across Salons, Auto Detailing, Bakeries & CSC
Live Sites: Salon Studio ↗ Bakery Shop ↗ Detailing ↗ CSC Portal ↗
Python Cloudflare Pages HTML5/CSS3 Wrangler Automation
Live Demos Code
📈
Data Analytics

Student Analytics

Multivariate EDA Group-wise academic performance metrics

Data Cleaning & Demographic Profiling

Comprehensive statistical investigation analyzing demographic, socio-economic, and study habits influence on academic success scores using Pandas and Seaborn.

Pandas NumPy Seaborn EDA
Code
🚢
ML Benchmark

Titanic ML & EDA

Feature Pipeline Survival pattern modeling & imputation

Advanced Feature Engineering Engine

Rigorous exploratory data analysis and feature engineering pipeline extracting title categories, deck positions, and family group survivability patterns.

Python Scikit-Learn Feature Engineering
Code
Technical Competencies

Skills & Tech Stack

Proficiencies across machine learning algorithms, deep learning, MLOps tooling, and software engineering.

🤖

AI & Machine Learning

Multi-Agent LLMs (CrewAI) 94%
XGBoost / LightGBM 95%
Explainable AI (SHAP) 92%
PyTorch / Computer Vision 88%
Scikit-Learn & SMOTE 95%
⚙️

MLOps & Cloud Serving

FastAPI / REST APIs 92%
Docker & Containerization 88%
DVC (Data Version Control) 85%
MLflow Experimentation 86%
Streamlit / Web UI 94%
📊

Data Analytics & Math

Python (Advanced) 96%
Pandas & NumPy 95%
EDA & Statistical Testing 90%
SQL & Data Modeling 85%
Matplotlib & Seaborn 90%
💻

Software & Cloud

Git & GitHub Actions 92%
Cloudflare Pages & Edge 90%
HTML5 / Modern CSS / JS 88%
Linux / Bash Scripting 87%
System Architecture 90%
Developer Playground

Interactive Recruiter CLI

Test drive my capabilities directly from the command line. Type commands or click the chips below!

rahul@ai-engine: ~ (zsh) v2.4.0
Welcome to Rahul Kumar Pradhan's interactive portfolio terminal! 🚀
Type 'help' to see available commands or click the quick chips below.
rahul@rkpcode:~$
whoami skills projects metrics education hire contact clear
Background & Growth

Academic & Engineering Journey

How structural engineering precision merged with behavioral media insights to form a distinctive AI engineering perspective.

AI & Engineering Milestones

2024 — PRESENT

AI & ML Systems Engineer

Autonomous AI & Open Source • Global

Architecting end-to-end machine learning pipelines, Explainable AI systems (SHAP in FinTech), and automated high-throughput web factory engines. Maintaining 18+ public GitHub repositories.

2023 — 2024

Machine Learning & Anomaly Research

Data Science Labs

Developed the Retail Margin Intel anomaly detection engine with SMOTE balancing, reducing 15+ hours/week of manual margin leak audits with 85.2% precision.

2022 — 2023

Computer Vision & Defect Segmentation

Industrial Deep Learning Project

Engineered pixel-level U-Net defect detection models in PyTorch for high-precision steel manufacturing quality assurance.

Academic Foundations

POST-GRADUATE

Master's Degree in Mass Media

Behavioral Insights & Communication

Applied psychology of audience engagement, communication dynamics, and textual narrative structures. Directly powers my intuitive approach to NLP and customer churn behavior modeling.

UNDERGRADUATE

Bachelor of Technology in Civil Engineering

Precision & Systems Foundation

Rigorous training in mathematics, structural integrity, quantitative tolerance analysis, and complex project execution under tight engineering constraints.

Ready to Deliver Value

Let's Build Something Extraordinary

Whether you're looking for a full-time AI/ML Engineer, an MLOps specialist to productionize your models, or an auditable predictive system for your team — I'd love to connect!

Primary Email (Click to copy)
contactrkp21@gmail.com
LinkedIn Network
linkedin.com/in/rkpcode
GitHub Portfolio
github.com/rkpcode
X / Twitter
@pradhanr776

Send a Direct Message

Email copied to clipboard!