My Academic & Achievements Journey

A timeline highlighting my education, hackathon achievements, and technical growth.

2023

Bachelor of Technology – Data Science

Manipal Academy of Higher Education, Udupi

· Expected Graduation: July 2027

Specialization

Artificial Intelligence, Machine Learning, and Data Science

Technical Skills

Hugging Face (Transformers)
Data Analysis (Classification, Prediction)
Deep Learning Models (CNN, RNN, Transformers)
Frontend & Backend Development
Database Management (PostgreSQL)
Programming Languages: C++, Python, Java

Relevant Coursework

Graph Theory
Number Theory
Calculus and Integration
Bayesian Probability
Machine Learning & Deep Learning (Attention, Transformers)
Database Systems (SQL, Oracle, PostgreSQL)
High-Performance Computing (CUDA)
Cloud Computing (Supabase, Firebase)
Data Structures and Algorithms
Object-Oriented Programming

2024

Developed an interactive mobile application in collaboration with Manipal Academy of Higher Education (MAHE) and KMC Hospital, Udupi, aimed at connecting cancer patients with their assigned doctors. The application facilitates both pre-operative and post-operative care, allowing patients to track their treatment and communicate efficiently with medical staff.

Pre-op and post-op calendar integration to display chemotherapy sessions, surgery dates, and follow-ups.
Emergency contact feature to reach doctors immediately in critical situations.
Comprehensive patient data collection including lifestyle habits (smoking, alcohol consumption), vitals tracking (e.g., blood pressure, BMI), medication reminders, dietary intake, and the ability to upload images of affected areas.
Secure authentication system for patients and KMC medical staff using Firebase Authentication.
Role-specific dashboards for doctors and patients to manage treatment plans and monitor progress.
Backend database storage implemented with Firebase Firestore for scalable and real-time data management.
Flutter frontend development leveraging packages such as `provider` for state management, `flutter_local_notifications` for reminders, and `image_picker` for media uploads.

Tools & Technologies: Flutter (Frontend), Firebase Authentication & Firestore (Backend), Figma (UI/UX Prototyping), Flutter Packages – Provider, Image Picker, Local Notifications.

Achievements: Successfully delivered a fully functional and interactive application to KMC Hospital, enabling seamless communication and care management for cancer patients. The project won a cash prize of ₹20,000.

GitHub: Cancer Gateway App

April 5, 2025

MongoDB: The Complete Guide to NoSQL Database Development

by EDUCBA

Verification

View Verification MongoDB Certificate

Oct 16, 2025

Course

Natural Language Processing in TensorFlow — DeepLearning.AI

Verification

View Verification NLP TensorFlow Certificate

October 27, 2025

Course

Generative AI with Large Language Models — DeepLearning.AI

Verification:

View Verification Generative AI with Large Language Models Certificate

My projects

Metal Health Therapy AI Agent

This is a mental health AI agent designed to provide support to individuals in need. It is built using LangChain for agent orchestration and prompt engineering, FastAPI to expose the backend endpoints, Twilio for emergency contact functionality, and Ollama to run models locally on the device. The current models in use are alibayram/medgemma:4b and Qwen2.5:7B. Future plans include fine-tuning the models, developing a frontend, and adding new features based on user feedback collected through the landing page. The web application will be launched after securing a suitable investor.

View Project

Landing page form my Metal Health Therapy AI Agent

this is the landing page for my Mental Health Therapy AI Agent project. It provides information about the project, its features, and how to get involved or support the development.

Explore Concept

Transformer_Text_Style_Prediction

This system implements a decoder-only Transformer language model inspired by Attention Is All You Need. The architecture consists of token embeddings, learned positional embeddings, and a stack of causal self-attention Transformer blocks. Each block contains multi-head masked self-attention, a position-wise feed-forward network, residual connections, and Layer Normalization. Causal masking is enforced using a lower-triangular attention mask to prevent information leakage from future tokens. The attention mechanism follows scaled dot-product attention, where queries, keys, and values are linearly projected from the embedding space. Multiple attention heads operate in parallel and are concatenated before a projection back to the model dimension. The output is passed through a linear language modeling head and optimized using cross-entropy loss for next-token prediction. The model is trained autoregressively on character-level input sequences, deviating from the original paper which uses subword tokenization. Positional encodings are learned embeddings rather than sinusoidal. The architecture omits an encoder stack, cross-attention layers, label smoothing, and learning rate warmup, making it simpler than the original Transformer. Dropout is configurable but currently disabled. Input and Output Input: A sequence of character indices with a fixed context window (block_size). Output: Logits over the character vocabulary and generated text via probabilistic sampling. This implementation prioritizes conceptual clarity and local inference over scale and optimization fidelity.

View Project

CANCER_GATEWAY_APP

this the same ap explained in the about

View Project

My Resume

↓

Download

Get in touch

Fill out the form and I’ll respond within 24 hours.

kosurusai646@gmail.com

Phone

+91 9515457049

Location

India