"From pure mathematics to applied AI — bridging theory and impact."
Senior educator & practitioner in AI, Data Science & Applied Mathematics, currently teaching graduate learners across industry-oriented programs. I turn mathematically heavy ideas into clear, practical intuition.
- 🔬 Research: Topological Data Analysis, Persistent Homology, Multi-Objective Optimization
- 🤖 Building: Agentic AI systems, RAG pipelines, ML tools & educational platforms
- 📐 Teaching: ML, DL, GenAI, RL, Time Series, MLOps — with hands-on labs & visual intuition
- 🏆 Best Paper Award at ICIVC 2021 (Oman) — TDA Feature Extraction
- 📦 Published:
timesage-tson PyPI — The Wise Time Series Library - 🌐 Website: mathcanvasmj.com — AI-Powered Visual Math & Science Whiteboard
- 🤗 Hugging Face: 10 Live AI Spaces — PRO member, building interactive ML apps
- 🏥 Healthcare AI: 4-Bot Clinical Suite — Diagnosis, Mental Health, Radiology, Anesthesia
|
An AI-powered visual whiteboard for mathematics and science education. Designed for interactive problem-solving, visual learning, and collaborative STEM exploration. • AI-assisted math problem solving • Interactive visual whiteboard |
Upload your resume and get roasted by AI with brutal honesty, humor, and actually useful career advice   
|
AI-powered digital twin of Dr. Milan Joshi — ask me anything about my research, teaching, or AI   
|
|
Explore ML models interactively — no code required. Tweak hyperparameters, visualize decision boundaries, compare algorithms   
|
Interactive Persistent Homology & Betti Numbers — visualize topological features of point cloud data   
|
|
AI Telegram bot for Deshpande Coaching Classes — instant doubt resolution & study assistance  
|
Personal AI assistant on Telegram — handles tasks, answers questions, and manages workflows  
|
|
AI-powered clinical diagnosis with voice stress analysis, drug interactions, document parsing & RAG   
|
Mental health support with 6 therapy modalities, voice stress, facial emotion AI & crisis detection  
|
|
Medical imaging analysis — X-ray, CT, MRI with DenseNet-121 pathology detection & GradCAM heatmaps   
|
World's first anesthesia AI — real-time vitals, pharmacokinetic dosing, ASA classification & Mallampati   
|
-10b981?style=for-the-badge&labelColor=0d1117){kind=icon}
A comprehensive 4-bot healthcare AI suite built entirely on free Hugging Face Spaces (CPU-only). Each bot features RAG-powered knowledge retrieval from WHO guidelines, clinical protocols, research papers & textbooks. No API keys needed. No GPU required. Fully open source.
An AI-powered clinical diagnosis assistant that conducts multi-turn medical interviews like a physician.
| Feature | Description |
|---|---|
| Chain-of-Thought Diagnosis | Differential diagnosis with step-by-step clinical reasoning |
| Urgency Triage | RED / YELLOW / GREEN severity scoring with actionable advice |
| Voice Stress Analysis | Record via microphone — extracts pitch, jitter, shimmer, speech rate to score stress 0-100 |
| Document Upload | Upload prescriptions, lab reports (PDF/image) — auto-extracts medications, lab values, diagnoses |
| Drug Interaction Checker | 100+ medication interactions with severity levels |
| RAG Knowledge Base | 50+ medical sources: WHO guidelines, Harrison's, Robbins, AHA/ACC, NICE, PubMed |
| Patient History Memory | Tracks symptoms, medications, vitals across the entire session |
How to use:
- Open the Clinical Chat tab and describe your symptoms naturally
- The bot asks follow-up questions like a doctor (age, duration, severity, medications)
- Switch to Voice Analysis tab — record a voice sample to detect stress biomarkers
- Switch to Document Upload tab — upload a prescription/lab report PDF or image
- Extracted medications are auto-checked for drug interactions
A mental health support bot with 6 evidence-based therapy modalities and multimodal emotion detection.
| Feature | Description |
|---|---|
| 6 Therapy Modalities | CBT, DBT, ACT, Motivational Interviewing, Mindfulness, Psychoeducation |
| Emotion Detection | Multi-dimensional sentiment analysis (valence + arousal + category) |
| Voice Stress Analysis | Microphone recording — detects anxiety markers via pitch, jitter, speech rate |
| Facial Expression AI | Webcam capture — 8-emotion classification (hsemotion + ONNX Runtime) |
| Multimodal Fusion | Combines text (40%) + voice (30%) + face (30%) for holistic mood scoring |
| PHQ-9 & GAD-7 | Validated clinical screening tools for depression and anxiety |
| Crisis Detection | Automatic crisis keyword detection with immediate safety resources (988, etc.) |
| Mood Tracking | Visual mood trend chart across the session with trend analysis |
| Journaling | AI-reflected journal entries with emotion analysis |
How to use:
- Open the Chat tab and share how you're feeling
- Type
PHQ-9orGAD-7to take validated depression/anxiety screenings - Type
journal Today I felt...to save a reflective journal entry - Switch to Voice Analysis tab — record your voice for stress biomarker detection
- Switch to Facial Analysis tab — use your webcam for real-time emotion detection
- Type
CBT,DBT,ACT, ormindfulnessto switch therapy modes
The first chatbot that can actually analyze medical images with AI-powered pathology detection.
| Feature | Description |
|---|---|
| Multi-Modality Support | Accepts X-ray, CT scan, and MRI image uploads |
| DenseNet-121 (CheXNet) | Detects 14 chest pathologies (pneumonia, cardiomegaly, effusion, etc.) |
| GradCAM Heatmaps | Visual overlay showing exactly where the AI is focusing |
| Natural Language Reports | AI-generated radiology reports in plain language |
| Confidence Scoring | Uncertainty quantification for each finding |
| Conversational Follow-up | Ask questions about the AI's findings |
| RAG Knowledge Base | Radiology guidelines, imaging protocols, pathology references |
How to use:
- Upload a medical image (X-ray, CT, or MRI)
- The AI analyzes the image and generates a detailed report
- View the GradCAM heatmap to see which regions the AI flagged
- Ask follow-up questions about the findings in the chat
World's first AI anesthesia assistant — no comparable tool exists in the market (as of 2026).
| Feature | Description |
|---|---|
| Real-Time Vitals Dashboard | Live simulation of HR, BP, SpO2, EtCO2, temperature with alerts |
| Pharmacokinetic Dosing | Marsh, Schnider & Minto models for propofol, remifentanil, fentanyl |
| ASA Classification | Physical Status scoring (ASA I-VI) based on patient history |
| Mallampati Assessment | Airway difficulty prediction (Class I-IV) |
| Drug Interaction Checker | Anesthetic agent interactions (propofol, sevoflurane, succinylcholine, etc.) |
| Pre-Op Risk Calculator | Cardiac, pulmonary, and bleeding risk stratification |
| RAG Knowledge Base | Miller's Anesthesia, ASA guidelines, WHO surgical safety protocols |
How to use:
- Start a chat and provide patient details (age, weight, medical history)
- The bot calculates ASA classification and pre-op risk scores
- Ask about drug dosing — get pharmacokinetic calculations for specific agents
- Use the vitals dashboard to simulate intraoperative monitoring scenarios
Architecture: All 4 bots run on free HF Spaces (2 vCPU, 16GB RAM, zero GPU). Built with Gradio 5.50, BM25 RAG retrieval, librosa (voice), hsemotion-onnx (face), OpenCV, pdfplumber, and pytesseract.
Topological Data Analysis ██████████████████░░ 90% Persistent Homology, Betti Numbers
Machine Learning & DL █████████████████░░░ 85% PyTorch, TensorFlow, scikit-learn
Time Series Analysis █████████████████░░░ 85% ARIMA, ETS, ML Forecasting
Reinforcement Learning ████████████████░░░░ 80% Policy Gradient, Q-Learning, PPO
Generative AI & LLMs ███████████████░░░░░ 75% RAG, Agents, Prompt Engineering
Multi-Objective Optimization ██████████████░░░░░░ 70% NSGA-II, MOALO, Pareto Fronts
Context Engineering █████████████░░░░░░░ 65% Token Management, Memory Systems
| # | Paper | Journal / Venue | IF |
|---|---|---|---|
| 1 | Comparison of NSGA-II, MOALO & MODA for Micro-Machining | Materials (2021) | 3.625 |
| 2 | Response Surface for Multi-Objective Optimization of Composites | Steel & Composite Structures (2019) | 3.89 |
| 3 | PSO-Tuned SVM Metamodels for Turbulent Flows | Engineering Computations (2020) | 1.246 |
| 4 | Persistent Homology for Big Data & Machine Intelligence | Springer AISC (2020) | — |
| 5 | TDA Feature Extraction 🏆 Best Paper Award | ICIVC-2021, Oman | — |
🤖 AI & Machine Learning (click to expand)
| Project | Description |
|---|---|
| 🤖 Ask My AI Clone | Chat with an AI trained on 15+ research papers — RAG + LangChain + GPT-4o |
| 📡 AI Research Radar | Self-updating arXiv paper tracker — GitHub Actions every 6 hours |
| 🧠 Awesome Agentic AI | 100+ agentic AI frameworks, papers, tools & resources |
| 🧬 Generative AI | LLM apps, prompt engineering, RAG systems & AI agents |
| 🔗 RAG Systems | Vector databases, embeddings, chunking & LLM integration |
| 💬 NLP | Text classification, sentiment analysis, NER, BERT fine-tuning |
| 🎮 Agentic AI | Tool use, multi-agent systems & LLM experiments |
| 🔬 AI-Science Projects | Computer Vision, LLM apps, multimodal AI prototypes |
📊 Data Science & Analytics (click to expand)
| Project | Description |
|---|---|
| 📈 Time Series Collection | 40+ univariate datasets with resources for forecasting |
| 📊 AI Job Market Dashboard | Interactive dashboard — salary trends, skills, geography |
| 📐 Loss Functions | Deep dive — MSE, Cross-Entropy, Focal, Huber & more |
| 🔍 Case Studies | End-to-end data science with real business problems |
| 📊 Data Analysis | EDA, statistics & visualization across diverse datasets |
| 🚀 MLOps | CI/CD pipelines, model registry & production deployment |
🎓 Education & Learning Resources (click to expand)
| Project | Description |
|---|---|
| 🎬 Animated Proof Gallery | Manim-animated proofs — TDA, Fixed Point, SVM, Pareto |
| 🎓 PhD Survival Kit | Paper writing, LaTeX, reviewer responses & Best Paper story |
| 📐 Math for AI | Linear algebra, calculus, probability for AI practitioners |
| 🧮 Linear Algebra for DS | Vectors, SVD, PCA with Python & visual intuitions |
| 🔢 ML From Scratch | 10 algorithms from scratch in NumPy with math |
| 📉 Optimization Guide | GD to Adam — benchmarks, visuals, notebooks |
| 🎮 RL Labs | Interactive Reinforcement Learning labs in browser |
| 📝 DS Cheatsheets | 11 cheatsheets — Python, SQL, ML, DL, Pandas |
🌍 AI Landscape & Research Curation (click to expand)
| Project | Description |
|---|---|
| 📚 Publications | Complete portfolio — 15 papers, DOI links, BibTeX |
| 🧠 Context Engineering | 34+ papers, 22+ videos, 27+ blogs, tools & roadmap |
| 🏥 AI Healthcare | Imaging, surgery, oncology, EHR, drug discovery |
| 🌏 Countrywise AI | 30+ countries — rankings, investment, talent, policy |
| 🔭 AI Landscape 2025 | Quantum computing, summits, tools, industry trends |
| 💡 Prompts | 500+ AI prompts — Manim, video, image, coding |
| 🔌 MCP Servers | 6 MCP servers for Claude — Manim, ML, Data Science |
| 🔨 AI Tools | Video, audio, PPT tools, YouTube, FAANG blogs |
| ⚡ n8n Workflows | 5,000+ workflows, 20 categories, 400+ integrations |
💎 Research & Apps (click to expand)
| Project | Description |
|---|---|
| 🩺 MedDiag-Pro | AI diagnosis bot — voice analysis, drug interactions, document parsing, RAG |
| 🧠 MindCare-AI | Mental health bot — 6 therapy modes, facial emotion AI, multimodal fusion |
| 📋 RadiScan-AI | Medical imaging — DenseNet-121 pathology detection, GradCAM heatmaps |
| 💉 AnestheBot-AI | World's first anesthesia AI — vitals dashboard, pharmacokinetic dosing |
| 🌀 TopoVista | Topological Data Analysis interactive application |
| 🎯 Manim Presentations | Beamer + Manim animated slides for lectures |
| 🎬 Video Analysis | Object detection, tracking & action recognition |
| 🌐 Web Apps | Full-stack TypeScript apps with modern frameworks |
| 🚀 Astro Starter | Modern web platform template with Astro & SSR |
Languages & Core
ML & AI
Tools & Platforms
✨ Baking mathematical intuition into AI systems since 2015 ✨