Highly accomplished MSc Computer Science graduate specializing in Artificial Intelligence, with a strong academic record (10/10 thesis, 105/110 overall grade) and a solid foundation in mathematics. Proven expertise in scientific machine learning, Physics-Informed Neural Networks (PINN), and IoT, demonstrated through impactful research, publications, and practical project development. Eager to apply advanced AI/ML skills to solve complex engineering and data-driven challenges in a dynamic research or industry environment.
Remote
→
Summary
Contributed to the development of flexible and scalable IoT remote sensing systems, optimizing data for commercial and highly dependent environments.
Highlights
Managed and developed firmware using Particle RTOS, ensuring robust and efficient operation of IoT devices across various deployments.
Implemented IoT Fleet management solutions, overseeing the deployment and maintenance of numerous remote sensing systems.
Integrated GIS solutions with IoT data, providing enhanced spatial analysis and data visualization capabilities for clients.
Managed agricultural IoT projects from conception to deployment, delivering data-driven insights for optimized farming practices.
Masters Thesis Student
Kiel, Schleswig-Holstein, Germany
→
Summary
Led scientific machine learning research focused on solving non-linear Partial Differential Equations (PDEs) for inverse engineering problems using Physics-Informed Neural Networks (PINN).
Highlights
Scaled Physics-Informed Neural Networks (PINN) for multi-scale problems by implementing transfer learning, significantly enhancing model adaptability and computational efficiency.
Trained PINN models on a 1D wave equation with diverse boundary and initial conditions, demonstrating robust predictive capabilities in complex physical simulations.
Utilized variable velocity models and Ricker Wavelet source terms to accurately simulate complex wave propagation scenarios, improving model realism.
Developed and trained PINNs for forward 2D Wave Equation problems, achieving high fidelity in simulations and contributing to a published journal article.
Optimized PINN performance through advanced techniques including temporal loss weighting, transfer learning, and self-attention mechanisms, improving convergence and accuracy by a measurable margin.
Successfully performed inversion of 2D wave under variable velocity models, validating model effectiveness in complex inverse problems.
Research Grant Holder
Genoa, Liguria, Italy
→
Summary
Conducted research under a grant to develop an Automatic Passenger Counting System, leveraging machine learning and IoT technologies.
Highlights
Designed and implemented a hardware setup using Raspberry Pi to accurately sniff Wi-Fi packets, forming the foundation for real-time data collection.
Developed a de-randomization algorithm utilizing Machine Learning to estimate passenger numbers with improved accuracy and efficiency.
Created and applied an algorithm based on Density-Based Clustering for efficient and reliable passenger counting, contributing to a published article.
→
Master of Science
Computer Science (Data Science and Engineering)
Grade: Thesis Grade: 10/10, Overall Grade: 105/110
Courses
Advanced Machine Learning
Digital Signal and Image Processing
Data Visualisation
Multi Agent Systems
Large Scale Computing
Virtualisation and Cloud Computing
Natural Language Processing
Speech Processing and Recognition
→
Bachelor of Science Honours
Mathematics
Grade: First Division with Honors
Courses
Mathematics
Statistics
Physics
Awarded By
University of Genova, Italy
Received a research grant to develop an Automatic Passenger Counting System in collaboration with Flairbit Sri and DIBRIS, University of Genova.
Awarded By
Babasaheb Bhimrao Ambedkar University (A Central University), Lucknow
Recognized with the Young Scientist Award at Springer's International Conference for Environmental Sustainability.
Awarded By
Aligarh Muslim University
Awarded second prize for presenting a research paper during bachelor's studies, showcasing early research aptitude.
Awarded By
Government of India
Secured first position at a 120-hour national hackathon, winning a cash prize of approximately 1000 USD for innovative hardware solutions.
Published by
MDPI Applied Sciences
Summary
Authored a journal publication detailing the application of transfer learning to scale Physics-Informed Neural Networks for high-frequency and multi-scale problems.
Published by
MDP Sensors. Sensors
Summary
Co-authored an article on developing an edge-based automatic passenger counting system using unsupervised clustering techniques.
Published by
National conference on Environmental sustainablility: Innovations, traslations dimensions and way forward.
Summary
Presented at a national conference on an adaptive sampling-based system for environmental monitoring, focusing on sustainable innovations.
Physics Informed Neural Networks (PINN), Deep Learning, Graph Neural Networks, Mathematical Modelling, Natural Language Processing, Speech Processing and Recognition, Advanced Machine Learning, Machine Learning.
Python, Pandas / Numpy, D3 JS, SQL / NoSQL, Data Visualisation.
Docker, Ansible, Virtualisation, gCloud / AWS, GIT.
InfluxDB / Telegraf, ElasticSearch / Kibana, Grafana.
IoT, Particle RTOS.
Digital Signal Processing, Dynamic filter design (Kalman).
PyTorch, TensorFlow, Hadoop, OpenMP and MPI, Hugging Face Transformers.
API (REST), Large Scale Computing.
Contact: Irosasco@mit.edu
Contact: frank.wuttke@ifg.uni-kiel.de
Contact: zarghaam.rizvi@uwaterloo.ca
→
Summary
Developed a sentiment analysis system for Indian news headlines, leveraging advanced BERT models and deep learning frameworks.
→
Summary
Optimized the N-Body problem simulation through parallel computing techniques, significantly improving computational speed.
→
Summary
Designed and implemented an automated system for application provisioning and cluster management on Docker Swarm using Ansible.
→
Summary
Explored and implemented adaptive noise cancellation methods as part of a Speech Processing and Recognition course.
→
Summary
Implemented a 2D Kalman Filter from scratch for real-time object tracking in video sequences as a final master's project.