ADITYA DAGA | 20NA30003

OCEAN & NAVAL ARCH / OCEAN ENGG. NAVAL ARCH. (M.Tech Dual 5Y)
MICRO SPL. in ARTIFICIAL INTELLIGENCE AND APPLICATIONS
EDUCATION
Year Degree/Exam Institute CGPA/Marks
2025 M.TECH Dual Degree 5Y IIT Kharagpur 8.29 / 10
2020 CBSE Delhi Public School, Bikaner 95.4%
2018 CBSE Shri Jain Public School, Bikaner 94.6%
INTERNSHIPS
Market Analyst Intern | Axxela Research & Analytics | Gurugram | PPO May '24 - Jul '24
Objective: To analyse global events, economic data and market trends to gauge sentiment, leverage risk management to maximize returns
• Traded fixed income interest rates derivatives across international markets, including Europe & US, and received a pre-placement offer
• Performed trendline & price analysis using Candlestick patterns, technical indicators such as Fibonacci retracement and Bollinger Bands
• Conducted fundamental macroeconomic analysis correlating interest rates with currencies, equities, banking and 1Y, 5Y, 10Y yield curves
• Devised dynamic hedging strategies to pinpoint profitable entry & exit points, backtested with 1500+ trades in 3M and 6M EURIBOR
AI & ML Research Intern | GenVR Labs | Remote May '23 - Sep '23
Objective: To leverage Generative AI technologies for innovative applications in 3D modeling, 3D reconstruction, and novel view synthesis
• Achieved 95% accuracy in efficiently converting a Neural Radiance Field (NeRF) to a 3D mesh using Positional Encoding and PyMCubes
• Implemented RealFusion generating a 3D Reconstruction of an object from a single image leveraging the existing 2-D Diffusion models
• Enhanced PSNR by 12% by integrating Zero 1-to-3, enabling the combination of arbitrary novel views into NeRF for better rendering
• Reduced loss to 0.0516 by effectively optimizing COLMAP's error resolution through the implementation of Instant-NGP and SDS loss
PROJECTS
Financial Statements Sentiment Analysis | Self Project Jul '24 - Aug '24
Objective: To determine the sentiment of given financial statements, discerning whether they exhibit a positive, negative, or neutral nature
• Achieved 86.22% accuracy using RoBERTa based model for financial text classification using PyTorch and Hugging Face Transformers
• Applied lemmatization, stemming, custom tokenization and TF-IDF vectorization with NLTK on 5000+ statements for text preprocessing
• Conducted Exploratory data analysis using Pandas Profiling, and efficient model training using the Hugging Face's Accelerate library
• Attained an F1 score of 0.85 by using a robust training pipeline with transfer learning, early stopping and learning rate scheduling
• Assessed classical models like Random Forest, XGBoost, SVM and Naive bayes with confusion matrix, obtained a modest 79% accuracy
Brain Tumor Detection | Self Project Jan '23 - Feb '23
Objective: To detect the presence of a tumor and classify the type of brain tumor identified using axial, coronal and sagittal MRI scans
• Evaluated multiple CNN architectures including ResNet101 and VGG19 with K-Fold Cross Validation, tuned hyperparameters with Optuna
• Achieved a maximum recall score of 88.47%, precision score of 87.85%, and an accuracy score of 88.06% on the validation dataset
• Built an API using Flask framework and integrated it with a Deep Learning based Web Application and deployed it on Heroku and Azure
Faults Prediction based on Severity | Bachelor's Thesis Project | Prof. Neeraj Kumar Goyal | IIT Kharagpur Jan '24 - Apr '24
• Performed Time series forecasting using neural networks to predict the number of critical, major and minor faults across software releases
• Tuned hyperparameters in ANN, bi-directional LSTM and used Autocorrelation function for identifying optimal lag values in ARIMA
• Detected anomalies based on reconstruction error, achieving an adjusted R-squared value of 0.894 using a finetuned LSTM model
Hospital Management System | Self Project Aug '23 - Oct '23
• Built a Django-based web application using MySQL, focusing on patient registration, appointment scheduling, and record maintenance
• Implemented robust patient data entry modules and automated storage of diagnostic results, ensuring seamless and secure data handling
• Developed a dashboard for doctors to access patient records, schedule appointments, and receive health updates, streamlining workflow
• Implemented SMTP-based email notifications to regularly update patients on diagnoses, enhancing communication and engagement
COMPETITION/CONFERENCE
Data Analytics | Silver | General Championship | IIT Kharagpur Feb '24 - Mar '24
• Created a Social Care Scorecard using SDOH, analyzing 85K census tracts with 322 features to suggest personalized actions based on risk
• Reduced dimensions by 66% with PCA, Improved risk classification accuracy by 31% with Semi-supervised learning & Bayesian Network
• Applied DBSCAN (epsilon 0.4) to identify 10 regional clusters, integrated RAG based LLM with ChromaDB for efficient feature selection
• Reduced validation loss by 20.76% and training loss to 0.012 by finetuning Llama-2-7b, Gemma-7b and Mistral-8x-7b with LoRA & QLoRA
SKILLS AND EXPERTISE
Programming Languages / Databases - C | C++ | Python | HTML | CSS | JavaScript | MySQL | MongoDB | MATLAB*
Frameworks / Libraries - Numpy | Pandas | Matplotlib | Sklearn | PyTorch | Tensorflow | NLTK | Node.js | Express.js | React.js | OpenCV*
Tools - Power BI | Tableau | Git and Github | MS Excel | Anaconda | VSCode | MaxSurf | Selenium | Azure* | ANSYS* | Docker* (*Familiar)

COURSEWORK INFORMATION
Curriculum: Programming and Data Structures | Artificial Intelligence:Foundation and Applications | Economics | Hydrostatics and Stability
Mathematics: Probability and Statistics | Stochastic Processes and Applications | Advanced Calculus | Linear Algebra | Transform Calculus
MOOCs: Data Structures and Algorithms in C++ | Machine Learning A-Z (Udemy) | Deep Learning Specialization by Andrew NG | Natural
Language Processing |Object Oriented Programming | Operating System | Database Management System | Financial Modelling & Valuation
EXTRA CURRICULAR ACTIVITIES
• Competitive Programming: Achieved a peak rating of 1469 (Specialist) on Codeforces [Adityakgpian01] and a rating of 1623 on Leetcode
• Sports: Captain of the Cricket team of Nehru Hall of Residence at the General Championship Sports & Games 2023-24, IIT Kharagpur
• Technology: Served as Secretary Technology, Nehru Hall of Residence, oversaw budgets & logistics for 5 tech events with 75+ members
• Contributions: Member of the Case Study, Chess and Silver winning Short Film-making team in the General Championship 2022-23
• Volunteering: Mentored students as part of the Student Welfare Group's mentorship program at IIT Kharagpur during the 2022-23 session
!Self declared by the student, CDC could not verify the relevant documents