Profile
Advancing Substructure Analysis in Tomograms
At the CMS experiment site, Characterising the VFAT3 electronics for GEM detectors.
Preaparing for the ME0 GE2/1-GEM MODULE Assembly for CERN.
Explaining the Clustering, ML techniques for 3D Tomographic data analysis.
In the Clean Room at INFN Bari during the GE2/1 GEM detector assembly.
Presenting density-based ML techniques at the IEEE METRO-XRAINE Conference at ST.Albans London, 2024.
Assembly and Quality Control of Sensor GE21
Projects Link
Browse my projects on GitHub →
Experience (Engineer & Researcher)
MLOPS ENGINEER & DATA SCIENTIST
at Independent Engineering Projects [October 2025 – Present]
Bari, Italy
- End-to-End MLOps Pipeline & Scalable API Deployment.
- Tech Stack: FastAPI, MLflow, Docker, Kubernetes (Minikube), AWS PostgreSQL, GitHub Actions.
- Model Tracking & CI/CD: Architected an automated continuous integration pipeline using GitHub Actions and integrated MLflow for rigorous experiment tracking, parameter logging, and model registry.
- Cloud Infrastructure: Deployed a centralized MLflow tracking server connected to an AWS PostgreSQL database, ensuring persistent, secure, and remote storage for model metrics and artifacts.
- Orchestration & Serving: Containerized a robust FastAPI inference application using Docker and orchestrated the complete system deployment on a Kubernetes (Minikube) cluster to ensure a highly scalable, fault-tolerant production environment.
Deployed AI Research projects and Web Applications
at my own website(www.mohrafik.it) [October 2025 – present]
Bari, Italy
- Developed and deployed a full-stack robust web application: Custom CMS & Security Dashboard.
- Tech Stack: Python, Flask, SQLAlchemy, MySQL/SQLite, Gunicorn, Railway, Integrated third-party APIs (Resend API).
- Deployed a robust, dynamic web application using Python and Flask, implementing a flexible SQLAlchemy ORM that intelligently transitions between local SQLite environments and production MySQL databases for scalable data management.
- Developed a custom Markdown-based Content Management System (CMS) and a live administrative dashboard featuring IP-API visitor geolocation, real-time navigation tracking, and advanced middleware security protocols to actively block malicious directory probing.
- Led the end-to-end cloud deployment lifecycle via Railway and Gunicorn, establishing automated third-party API pipelines (Resend API) for secure, instant email communication and advanced spam prevention.
Researcher
at STIIMA-CNR (Institute of Intelligent Industrial Systems and Technologies for Advanced Manufacturing) [May 2023 – October 2025]
Bari, Italy
- Deep Learning: Computer Vision & Digit Recognition.
- Tech Stack: Python, PyTorch/TensorFlow, NumPy, OpenCV.
- Designed and trained a custom Deep Neural Network (CNN/DNN) architecture for high-precision image classification using the benchmark MNIST dataset.
- Built an end-to-end automated data pipeline in Python to handle image preprocessing, normalization, and batch generation for training and evaluation phases.
- Executed advanced hyperparameter tuning (e.g., learning rate optimization, dropout regularization) and architectural refinements to prevent overfitting, achieving a classification accuracy exceeding 98%.
- Developing semi-automatic Python pipelines to segment 3D point-cloud and volumetric data for hidden substructure discovery.
- Building Flask-based APIs that connect Unity3D AR/VR clients to Python back-end clustering and reconstruction modules.
- Automating analysis and modelling of AFM protein spectroscopy data, including feature detection, outlier rejection and ML-based fitting.
Researcher
at National Institute of Nuclear Physics (INFN) Bari – CMS Experiment at CERN [Mar 2021 – Mar 2023]
Bari, Italy
- Developed and tested front‑end electronics and data acquisition systems for the CERN CMS experiment (GEM detectors).
- Characterized custom ASIC chips(VFAT3b) using a Modular Acquisition Board (MOSAIC) and analyzed the data for performance (e.g., calibration, optimalDAC settings, channel diagnostics) to ensure reliable operation.
- Led quality control for next‑generation particle detectors (GEM chambers for the CMS GE2/1 upgrade).
- Automated test procedures using LabVIEW and microcontroller‑based tools (Arduino). Conducted comprehensive QC tests (power/HV tests,gain uniformity, dead channel checks), improving the efficiency and accuracy of detector validation.
- Automated Raman peak detection, feature extraction, analysis, and visualization of extracted information from data files.
Research Engineer
at University of Delhi, Department of Physics [Mar 2015 – Mar 2020]
New Delhi, India
- Developed digital signal‑processing algorithms in MATLAB/C to reduce noise in detector signals and extract features (rise time, pulse width,etc.), significantly improving signal‑to‑noise ratio and particle detection accuracy. Also optimized FPGA firmware (Verilog/VHDL) for the detector readout electronics, enhancing data acquisition throughput.
- Contributed to R&D on advanced particle detectors (Resistive Plate Chambers and GEM detectors) for the CMS experiment and the India‑based Neutrino Observatory (INO).Assembled and calibrated detector modules, performing key tests (gas leakage, gain uniformity, efficiency measurements) to meet experimental specifications in collaboration with international teams (CERN CMS Upgrade project).
- Designed and implemented automation tools for detector testing and monitoring – including LabVIEW interfaces for high‑voltage control and current monitoring, and Arduino‑based modules for environmental sensing – streamlining data collection and quality assurance. (Co‑authored multiple research publications based on this work.
Education
Back to TopIndian Institute of Technology (IIT BHU)
Varanasi, India
Master in Technology (M.Tech), Digital Techniques and Instrumentation
2012 - 2014
- First Division (GPA 7.92/10).
- Thesis Title: Recognition and Analysis of Facial Expressions using Support Vector Machine (95% accuracy).
IETE (Institute of Electronics and Telecommunication Engineers)
New Delhi, India
Bachelor of Technology (B.Tech), Electronics & Telecommunication Engineering
2009
- Graduated with 7.18/10 CGPA (First Division).
Academic Achievements, Awards and Honors
2012
Excellent Score 99.36 percentile with All India Rank :1127 , Graduate Aptitude Test in Engineering (GATE), MHRD, Government of India
India
2012–2014
Awarded Scholarship for Master in Technology(M.Tech) Program, by MHRD, Government of India
New Delhi, India
Certificates
2026
Advanced Learning Algorithms (Machine Learning Specialization), DeepLearning.AI (Andrew Ng) and Stanford Online
2026
Supervised Machine Learning: Regression and Classification, DeepLearning.AI (Andrew Ng) and Stanford Online
2026
Tableau : Data to Dashboard Mastering Visual Storytelling with Tableau, NPTEL (IIT Madras)
2018
Certification in C and C++ programming with A1 grade, National Small Industries Corporation (NSIC)
New Delhi India
2017
Course on computer concepts (CCC), National Institute of Electronics and Information Technology (NIELIT) Ministry of Electronics and Information Technology Govt. of India
New Delhi India
Presentation
Back to TopIEEE International Conference on Metrology for eXtended Reality, Artificial Intelligence and Neural Engineering
ST ALBANS - LONDON, UK
PRESENTER FOR < Advancing Substructure Analysis in Tomograms >
OCTOBER 21-23, 2024
- Introduced tomography and its challenges in revealing hidden structures in 3D point cloud data.
- Introduced techniques for segmenting and advancing substructuring, and visualizing them immersively using extended reality and Unity.
Interests
- Reading
- Traveling
Professional Skills
- MLOps & Cloud Infrastructure: Docker, Kubernetes (K8s), CI/CD Pipelines, MLflow, Containerization, Linux, Git, AWS
- Data Analysis & ML: NumPy, Pandas, Matplotlib, Seaborn, scikit-learn, statistics
- Deep learning: PyTorch, TensorFlow, Keras
- 3D / Tomography: Open3D, volumetric data processing, point-cloud segmentation
- Data acquisition & control: Flask APIs, sensors, LabVIEW, DAQ systems
- Computer vision & signal processing: OpenCV, time-series analysis
- Hardware Implementation : FPGA (Verilog/VHDL), front-end electronics, microcontrollers
Tools & Technologies
- Python, C/C++, MATLAB, SQL, LaTeX
- Git, Linux, Unity (AR/VR), web scraping