CHRISTOPHER SCULLY
Queens, NY 347-***-**** adzjgt@r.postjobfree.com https://www.linkedin.com/in/christopher-scully-a8791920a/
https://codeforces.com/profile/AI-Master/
An accomplished Machine Learning Engineer with an impressive 7-year history of leading a variety of projects, including trailblazing information extraction from unprocessed images. Skilled in harnessing C++ algorithm proficiency developed through involvement in global programming competitions, particularly Codeforces. Continuously delivering inventive ideas, autonomous research, and outcomes while meeting demanding timelines. Actively collaborating with interdisciplinary teams, collaborating with more than 10 developers and specialists in product management to steer project triumph
EXPERIENCE
MAR 2022 -
MAY 2023
SR. MACHINE LEANING ENIGNEER, AMAZON
-Developed a Convolutional Neural Network (CNN)-based Text Detection model as an enhancement to the OCR system, resulting in a remarkable 41% reduction in inference time for extracting vital information from scanned documents and card images.
-Enhanced extraction accuracy by 7% by combining multiple GCN-based KIE models into BERT-based Large Language Model (LLM) model.
-Devised an innovative algorithm for generating synthetic training data using OpenCV, skimage, and scipy, resulting in improved KIE model performance.
-Applied transfer learning techniques to train models based on cutting-edge pre-trained architectures.
-Selected the lifecycle for the named entity recognition project, and managed it using JetBrains IDEs by integrating MLflow and AML.
-Created a C# testing application, which shortened data analysis projection time by one week.
-Designed a C# application to efficiently prepare training data from real images.
-Managed the coding of over 1500 lines in Python and C++ within a Linux environment.
-Analyzed more than 1300 documents and card images to evaluate product accuracy.
-Offered guidance and mentorship to junior developers to ensure the smooth maintenance of projects.
JUL 2020 - JAN 2022
MACHINE LEARNING ENGINEER, META
-Championed a solution proposal to address a prominent customer complaint, resulting in the completion of the project six months ahead of schedule.
-Enhanced the performance of an automatic data processing machine through meticulous analysis, testing, and debugging of over 2500 lines of code.
-Introduced an innovative CNN-based method for detecting text and tables using Computer Vision, leading to a 54% reduction in inference time for scanned certificate documents and PDFs.
-Developed the project using C++ within a Linux environment, performing CUDA/OpenMP programming.
-Created a performance visualization tool using C# to monitor the outcomes of the detection model.
-Utilized both Pandas and PySpark as integral components in coordinating distributed computing processes.
-Simplified the process of managing infrastructure by Terraform, providing a consistent and repeatable way to create, update, and delete resources across various platforms.
JUN 2018 -
DEC 2019
MACHINE LEANING ENIGNEER, EPHESOFT
-Enhanced the efficiency of an automatic data processing machine through thorough analysis, testing, and debugging of over 3000 lines of code.
-Developed a denoising model that achieved an 11% performance improvement by removing noise and rain from scene images.
-Employing Azure, the real-time car number plate extraction project was developed and trained, followed by rigorous testing to evaluate its accuracy.
-Engaged in pioneering research on dynamic planning strategies to tackle traffic challenges.
-Executed the implementation of advanced denoising and deraining algorithms, transitioning from VGG-16 to MobileNetV3 to reduce processing time.
-Demonstrated proficiency in CNN-based deep neural networks and attention mechanisms using Caffe2 library.
SEP 2016 -
APR 2018
MACHINE LEANING INTERN, DATAEDGE
-Performed benchmarking on cutting-edge logo detection models and evaluated their performance using TensorFlow.
-Utilized Terraform to set up and oversee the necessary infrastructure for a logo detection project, guaranteeing a uniform and automated deployment process.
-Developed synthetic training data for logo detection using C++, OpenCV, and data augmentation methods.
-Proficiently debugged and optimized over 1000 lines of code.
-Demonstrated skill in utilizing OpenCV's image processing functions and the PyTorch library in Python.
-Contributed to the development of a deep learning model for detecting logos in construction processes, leading to a 30% reduction in downtime.
AI SKILLS
TECHNICAL SKILLS
Image Segmentation
Text Detection
OCR (Optical Character Recognition)
Named Entity Recognition
LLM
Image Denoising
Object Detection
Computer Vision
Image Processing
Data Generation
Model Creation
Code Analysis.
CUDA(GPU)
Python, C++, C#
PyTorch
OpenCV
NumPy
Caffe2
Skimage
Scipy
Pandas
PySpark
Scikit
TensorFlow
Keras
AWS, Docker
ACTIVITIES
The international programming contest – Codeforces contestant
Level: Div 1
https://codeforces.com/profile/AI-Master/