class Mukilan:
def __init__(self):
self.name = "Mukilan S"
self.location = "Chennai, Tamil Nadu 🇮🇳"
self.degree = "B.Tech – AI & ML @ REC, Chennai (2028)"
self.email = "mukilans25361@gmail.com"
self.public_repos = 19
self.stack = [
"Python", "TypeScript", "Java", "C",
"React", "Next.js", "FastAPI", "Vite",
"PyTorch", "TensorFlow", "Keras",
"Scikit-Learn", "XGBoost", "Gradio",
"LangGraph", "RAG", "Gemini",
"Firebase", "Supabase", "Docker"
]
self.currently_learning = [
"Transformers & Vision Models",
"LLM Fine-Tuning & RAG Pipelines",
"MLOps & Model Deployment",
"Agentic AI – LangGraph & Gemini",
"Azure AI Services"
]
self.fun_fact = "Predicted both 2022 WC finalists with my ML model! ⚽"
def motto(self):
return "Build. Learn. Deploy. Repeat. 🚀"
🤖 Internship in Generative AI — Altruisty Innovation Pvt Ltd | February 2026
Generative AILLMsAI Workflows
- 🚀 Completed an intensive 15-day internship focused on Generative AI technologies.
- 🧠 Gained hands-on experience and deep insights into the rapidly evolving field of GenAI and large language models.
🔧 PumpDiagnosisChallenge — Computer Vision Engineer | June 2026
PythonPyTorchResNet18Transfer LearningSTFTComputer VisionPredictive Maintenance
- 🏗️ Built a pump fault diagnosis system on the MANGO Platform using Transfer Learning with ResNet18 on STFT Spectrogram images.
- 🔬 Converted raw vibration signals into frequency-domain spectrograms and trained a fine-tuned CNN for multi-class fault classification.
- 🚀 Applied Transfer Learning to achieve high accuracy on limited industrial sensor data.
🔍 Defect-Detection — Anomaly Detection Engineer | June 2026
PythonPyTorchPatchCoreWideResNet50Unsupervised LearningComputer Vision
- 🏗️ Implemented a state-of-the-art anomaly detection system for industrial manufacturing using PatchCore algorithm with WideResNet50 backbone.
- 🎯 Achieved near-perfect detection without any defective training samples using unsupervised learning.
- 🔬 Extracted deep feature embeddings and built a coreset memory bank for efficient nearest-neighbor anomaly scoring.
🚨 StampedeZero — Edge AI Platform | July 2026
PythonPyTorchYOLOv8FastAPIReactComputer VisionEdge AI
- 🧠 Architected a real-time crowd safety platform using YOLOv8 for crowd density estimation and early stampede detection.
- 📡 Built a FastAPI + React dashboard with live video feed processing and alert systems.
- ⚡ Optimized for edge AI deployment with low-latency inference for real-time surveillance scenarios.
🌿 Ecosphere Management Platform — Full-Stack Developer | July 2026
TypeScriptNext.jsReactSupabaseESGGamificationSustainability
- 🌍 Built a comprehensive ESG (Environmental, Social, Governance) platform to monitor sustainability metrics and ensure compliance.
- 🎮 Implemented a gamification engine to incentivize employee green initiatives with points and leaderboards.
- 📊 Designed beautiful data visualizations and reporting dashboards for ESG insights.
- 🔗 Live: ecosphere-management-platform.vercel.app
🧠 Industrial-Operations-Brain — Agentic AI Engineer | June 2026
PythonLangGraphGeminiRAGKnowledge GraphFastAPINext.jsIoT
- 🤖 Architected an Agentic AI system fusing RAG + Knowledge Graph + Live IoT data for industrial knowledge intelligence.
- 🔗 Used LangGraph for multi-agent orchestration and Google Gemini as the core reasoning engine.
- 📡 Integrated real-time IoT sensor streams into the knowledge pipeline for adaptive industrial decision-making.
🤖 CompileAI — AI Pipeline Engineer | June 2026
PythonFastAPINext.jsPydanticOpenAICode Generation
- 🏗️ Engineered a 7-stage pipeline that converts natural language product requirements into validated, executable application specifications (UI, API, DB, Auth).
- 🔗 Live: compile-ai-rho.vercel.app
⚽ FIFA 2026 World Cup Predictive Analytics Engine — Project Lead | June 2026
PythonScikit-LearnXGBoostRandom ForestLogistic RegressionReactViteFirebaseMonte Carlo
- 🏗️ Architected a complete ML pipeline using Python & Scikit-Learn to predict World Cup match outcomes from historical data and team Elo ratings.
- 🎯 Trained ensemble models (Logistic Regression, Random Forest, XGBoost) and ran 10,000 Monte Carlo tournament simulations to compute win probabilities.
- ✅ Validated model accuracy by correctly predicting both 2022 FIFA World Cup finalists using historical data.
- 🌐 Built a React + Vite interactive web application and deployed globally via Firebase Hosting for zero-friction accessibility.
🌦️ AeroWeather — Frontend Developer | March 2026
React 18ViteGitHub ActionsREST APIsGlassmorphic UIResponsive Design
- 🚀 Engineered a high-performance weather dashboard using React 18 + Vite, integrating real-time weather and 7-day forecasting APIs.
- 📊 Built interactive radar maps and atmospheric analytics visualizations with a responsive glassmorphic dark-mode UI.
- ⚡ Boosted application load performance by ~40% through Vite build optimizations and lazy-loading strategies.
- 🔄 Automated deployment pipelines with GitHub Actions, achieving 100% CI/CD coverage and eliminating manual release steps.
🔢 MNIST ANN Digit Classifier — Project Lead | December 2025
PythonTensorFlowKerasGradioGitHub ActionsPytestRuffCI/CD
- 🧠 Designed and trained ANN architectures with Dropout and Early Stopping regularization for MNIST handwritten digit classification.
- 🏆 Achieved 98.08% test accuracy with an F1-score of 0.98 across all 10 digit classes on 10,000 test images.
- 📐 Benchmarked against a CNN baseline achieving 99.07% accuracy, producing a thorough comparative model analysis.
- 🔁 Established a full CI/CD pipeline using GitHub Actions, Pytest, and Ruff — 100% automated validation on every push.
| 🗂️ Project | 🛠️ Stack | ✨ Highlights & Metrics |
|---|---|---|
| 🔧 PumpDiagnosisChallenge | Python · PyTorch · ResNet18 · STFT | Transfer learning on spectrograms · Predictive maintenance |
| 🔍 Defect-Detection | Python · PyTorch · PatchCore · WideResNet50 | Zero-shot anomaly detection · Unsupervised learning |
| 🚨 StampedeZero | Python · YOLOv8 · FastAPI · React | Real-time crowd safety · Edge AI |
| 🌿 Ecosphere Platform | TypeScript · Next.js · Supabase | ESG monitoring · Gamification · Live 🔗 |
| 🧠 Industrial-Ops-Brain | Python · LangGraph · Gemini · RAG | Agentic AI · Knowledge Graph · IoT fusion |
| 🤖 CompileAI | Python · FastAPI · Next.js · OpenAI | NL → App specs · 7-stage pipeline · Live 🔗 |
| ⚽ FIFA 2026 Predictor | Python · Scikit-Learn · XGBoost · React · Firebase | 10,000 simulations · Predicted 2022 WC finalists |
| 🌦️ AeroWeather | React 18 · Vite · REST APIs · GitHub Actions | ~40% faster load · Full CI/CD · Live 🔗 |
| 🔢 MNIST Digit Classifier | Python · TensorFlow · Keras · Gradio · Pytest | 98.08% accuracy · 0.98 F1-score · 100% CI/CD |
| 🏅 | Achievement | Details |
|---|---|---|
| 🏆 | Winner - MANGO Platform AI/ML Workshop | Organized by Walmart Center for Tech Excellence, IIT Madras (June 2026) |
| 🏅 | Applied Machine Learning and AI | CII, Centre for AI |
| 🏅 | Machine Learning & Deep Learning Onramp | MathWorks |
| 🏅 | Data Analytics & Cyber Job Simulations | Deloitte |
| 🤖 | Microsoft Learn – AI & Gen AI Basics | Certified by Microsoft Learn |
| 🪁 | Microsoft Learn – Microsoft Copilot | Get Started with Microsoft Copilot certification |
| 🗄️ | LinkedIn Learning – Database Management | Database Foundations: Database Management Certification |
| 🧠 | End-to-End AI Projects | Delivered 9+ production-grade AI/ML projects independently |
| 🔬 | Industrial Computer Vision | PatchCore & Transfer Learning for manufacturing defect detection |
| 🤖 | Agentic AI Systems | LangGraph + Gemini multi-agent RAG + IoT fusion system |
| ⚽ | FIFA 2022 WC Prediction Validation | Model correctly predicted both 2022 finalists from historical data |
| 🎯 | MNIST 98.08% Accuracy | ANN model achieving 0.98 F1-score across 10,000 test images |
| 🔥 | 40% Performance Improvement | AeroWeather load time reduction via Vite optimization |
| 🎓 Degree | 🏫 Institution | 📅 Year | 📊 Score |
|---|---|---|---|
| B.Tech – Artificial Intelligence & Machine Learning | Rajalakshmi Engineering College, Chennai | Expected 2028 | In Progress |
🤖 Agentic AI → LangGraph, Multi-Agent Systems, Tool Use, Gemini APIs
🧱 Generative AI → LLM Fine-Tuning, Prompt Engineering, RAG Pipelines
🔬 Deep Learning → Transformers, Attention Mechanisms, Vision Models
☁️ MLOps → Model Serving, Docker, CI/CD for ML, Azure ML
📊 Advanced ML → Ensemble Methods, Bayesian Optimization, AutoML
🌐 Full-Stack AI → FastAPI, Next.js, WebSockets, Real-Time Inference
⭐ If you find my work interesting, consider giving a star to my repos!
Let's connect and build something amazing together 🚀