PKCode — Where Code Meets Curiosity
I'm a passionate Software Engineering student with a strong foundation in Artificial Intelligence, Agentic AI systems, and MLOps.
┌────────────────────────────────────────┐
| │
| ██████╗ ██████╗ ██████╗ ███████╗ │
| ██╔════╝ ██╔═══██╗██╔═══██╝██╔════╝ │
| ██║ ██║ ██║██║ ██╗█████╗ │
| ██║ ██║ ██║██║ ██╗██╔══╝ │
| ╚██████╗ ╚██████╔╝╚██████╗╚███████╗ │
| ╚═════╝ ╚═════╝ ╚═════╝ ╚══════╝ │
| │
| > experiments loaded: 10 │
| > status: forging │
| > last spark: today │
| │
└────────────────────────────────────────┘
Artifacts
AI-powered mobile nutrition assistant that analyzes food images for calorie and macronutrient estimation. Combines ML-based meal recognition with FastAPI backend inference and personalized dietary insights.
Ensemble of ResNet, DenseNet, and EfficientNet to detect diabetic retinopathy. Deployed via Flask. MLOps ready.
Interactive Python application for visualizing Formula 1 telemetry and replaying race events. Integrates real-time telemetry parsing with high-performance graphical rendering.
Developed a plant pathology classification model with extensive augmentation (+40% dataset expansion). Integrated Grad-CAM heatmaps to enhance agricultural decision transparency and trust.
Built a carbon footprint estimation platform predicting CO₂e emissions per dollar spent using emission factor modeling. Designed for sustainable supply-chain analytics with full-stack deployment on Netlify.
Implemented advanced object segmentation using Segment Anything Model 2 (SAM 2) with multi-prompt inference support. Enables precise mask generation via point and bounding-box guidance.
Designed a transfer-learning-based CNN for early-stage liver fibrosis detection from ultrasound scans. Integrated focal loss for class imbalance and Grad-CAM for interpretable diagnostic visualizations.
A React app powered by Gemini API to convert code from one language to another. This project is made just to learn react and not for production.
Field Notes
Brief observations, technical findings, and thoughts on interesting research topics.
A paradigm shift where AI is deeply integrated into operating systems (OS), transforming intelligent agents and user interactions.
Implementing Model Context Protocol for seamless AI model interactions with vector databases in RAG apps.
Exploring how Segment Anything behaves on ultrasound and retinal scans, and its implications for volumetric image reconstruction
Running Llama2 locally and building a personal chatbot API for natural language tasks.
Winning as a Habit
Active experimentations with life. Things that I have won and learned from... so far.
Presented a technical paper at a national-level competition and secured 1st place.
Selected among 100 colleges nationwide (10,000+ delegates) to represent MIT Mysore as the sole Mysuru qualifier at India's largest startup congregation.
Won 1st place at 24-hour national hackathon against 120+ teams by engineering and optimizing a breast cancer detection model under extreme time constraints.
Won 2nd place in an inter-college Ideathon where we pitched a RAG system with MCP support.