class SaiMrudula:
institution = "Amrita Vishwa Vidyapeetham · B.Tech CSE (AI)"
location = "India"
focus = ["Applied ML", "Computer Vision", "Cybersecurity Analytics", "Biomedical Signals", "Conversational AI"]
looking_for = ["AI/ML internships", "research collaborations", "teams building useful systems"]
def build_style(self):
return "prototype fast, measure honestly, explain clearly"I build applied AI systems that combine engineering, experimentation, and readable documentation. My strongest work sits across smart-grid cybersecurity, markerless motion capture, EEG reconstruction, language-learning AI, and computer-vision workflows.
| Project | What it is | Stack / Focus |
|---|---|---|
| Smart Grid Cyberattack Detection | Intrusion-detection pipeline for smart-grid/IEC-104 traffic with ML benchmarking and publication-style outputs | Python, Scikit-learn, cybersecurity analytics |
| 3D Motion Capture | Markerless motion-capture system with pose tracking, stabilization, offline validation, and 3D visualization | Python, OpenCV, MediaPipe, Panda3D, React |
| MisSpoke | Voice-first AI language-learning platform with tutor personalities, translation, memory, and learning analytics | Next.js, React, TypeScript, ElevenLabs, Supabase |
| EEG Reconstruction with ADMM | Graph-signal-processing approach for reconstructing missing EEG data using local graphs and ADMM | Python, NumPy, SciPy, MNE |
| Cancer Driver Gene Classification | Bioinformatics ML pipeline for classifying cancer driver genes from structured biological features | Python, Pandas, Scikit-learn |
| Solar Panel Fault Detection | CNN-based computer-vision workflow for identifying visual faults in solar panels | Python, CNNs, image processing |
| AI Greenhouse | IoT desktop app for plant monitoring, Arduino sensor input, analytics, and automation | Python, PyQt5, Arduino |
| Double DQN Tic-Tac-Toe | Reinforcement-learning project using Double DQN to train a game-playing agent | Python, RL, neural networks |
| Project | Repo | Paper / Reference | Demo / Deployment | Dataset / Results |
|---|---|---|---|---|
| Smart Grid IDS | Repo | IEEE reference | Demo video: not uploaded yet | IEC-104 smart-grid traffic; result plots in results/, IEEE_Plots/, and Graphs/ |
| Motion Capture | Repo | Research/prototype project | Demo video: add to repo releases or README when finalized | Offline videos/webcam streams; diagnostics in analysis_results/ |
| MisSpoke | Repo | ElevenLabs Challenge build | Deployment link: add after hosting | User-session data stored through Supabase/IndexedDB; screenshots pending |
| EEG ADMM | Repo | Local graph + ADMM reconstruction study | Demo video: not uploaded yet | EEG-128channels ERP Lanzhou dataset; result images in images/ |
| EEG SVD | Repo | SVD EEG paper DOI | Demo video: not uploaded yet | Simulated EEG/EOG mixtures; SNR results documented in README |
| Portfolio Site | Repo | Personal portfolio project | Deployment link: add after GitHub Pages/Vercel deployment | Screenshots pending |
Languages
AI / ML / Vision
Web / Apps / Infra
- Biomedical signals: EEG acquisition, EEG SVD denoising, EEG ADMM reconstruction
- Computer vision: aircraft detection, solar fault detection, markerless motion capture
- Chatbots and language AI: Rasa bot, Language Tutor Bot, MisSpoke, recipe chatbot
- Systems and fundamentals: C file management with syscalls, OSI model simulation, microservices research
- Frontend/UI: portfolio-site, SchoolMate UI, mood writing assistant
AI Chatbots Intern · Academy of Robotics Pvt. Ltd.
Visakhapatnam, India · Jul 2024 – Sep 2024
- Built Rasa conversational agents from scratch
- Worked on intent classification, entity recognition, and NLU flows
- Designed prompt-engineered dialogue interactions
- Shipped working chatbot prototypes