name : R. Abishek
degree : Integrated M.Tech — CSE (Business Analytics), VIT Chennai
cgpa : 8.86 / 10
certified : AWS Cloud Practitioner ☁️
published : Frontiers Journal — Multimodal Parkinson's Detection with XAI
experience :
- Junior ML Engineer @ Omdena Bhutan Chapter (Jun–Aug 2025)
- Digital Transform Lead @ Kalai Vrikshya Architects (May–Sep 2025)
- Web Developer Intern @ Creatah Software Technologies (May–Jul 2024)
interests : Graph Neural Networks · XAI · Multimodal AI · MLOps · Healthcare AI
philosophy : "Precision over noise. Systems over scripts. Impact over impressions."| Project | What it does | Stack |
|---|---|---|
| 🧠 Multimodal Parkinson's + XAI | CNN · LSTM · ResNet50 fusion across speech, gait & handwriting. +18% over unimodal baselines. SHAP + Grad-CAM explainability. Published — Frontiers Journal. | PyTorch · XGBoost · SHAP |
| 🔬 Graph Drug Repurposing | Heterogeneous biomedical knowledge graph (Drug–Protein–Disease) + HeteroGraphSAGE for supervised link prediction. 84% F1. | Neo4j · PyG · SciSpaCy |
| 🔍 Dark Pattern Detector | Fine-tuned BERT across 8 deceptive UI categories. Real-time Streamlit dashboard + Selenium scraping. 97.5% accuracy. | BERT · Selenium · Streamlit |
| 🕸️ Fake Engagement Graph | Behavioural interaction graph to surface inauthentic engagement patterns at scale. | Graph ML · Python |
| 🏥 Chronic Care Risk Engine | 90-day deterioration forecast for heart failure, diabetes & COPD — with full XAI layer. | Ensemble ML · Calibration · XAI |
| 📄 IntelliDoc RAG | Multi-document retrieval-augmented generation — natural language Q&A across PDF stacks. | RAG · LangChain · PyTorch |
| 🧬 Gene Expression Analysis | Mitochondrial dysfunction + gene expression pipeline for Parkinson's research. | Python · Bioinformatics |
| 🔐 Anomaly-Based IDS | Network intrusion detection via ML/DL anomaly modelling on real traffic data. | ML · DL · Python |
| 🧩 Graph Resume Matcher | Knowledge graph + semantic embeddings to match candidates to roles with full context. | Neo4j · FastAPI · Embeddings |
| 🗣️ Attention TTS | Neural text-to-speech with Tacotron2 + HiFi-GAN and live attention visualisation. | Tacotron2 · HiFi-GAN · PyTorch |
| 💡 Emotion-Aware Smart Lamp | Real-time CV emotion + drowsiness detection piped to IoT-controlled adaptive lighting. | OpenCV · IoT · TTS |
| 📋 Resume Classifier | End-to-end NLP pipeline classifying PDF resumes — containerised and live on Render. | FastAPI · Streamlit · Docker |
Languages
ML / DL / AI
APIs / Web / DevOps / Cloud
Databases & Graph
class Abishek:
focus = "Multimodal XAI systems built for clinical trust — not just benchmark scores."
learning = ["Graph Neural Networks (advanced)", "LLM fine-tuning", "Production MLOps"]
open_to = ["Research collaborations", "AI/ML internships", "Genuinely hard problems"]
@staticmethod
def philosophy() -> str:
return (
"Build. Break. Understand. Improve. "
"That's how meaningful technology is created."
)Precision over noise. Systems over scripts. Impact over impressions.