$ whoami
Ted Soldatov
Comp Sci + Math Major, Software Engineering Mentor, Independent Software Developer
$ skills
ReactNode.jsPythonC++DockerKubernetes
System Architecture & Projects
American Sign Language (ASL) Recognition System
Built and trained a deep learning model capable of translating ASL hand symbols into text in real-time using a live video feed.
Key Achievements:
- Processed 55,500 images from Kaggle dataset, converting them to grayscale and normalizing pixel values through a custom Python preprocessing pipeline
- Split dataset into 70% training (38,850 images), 15% validation (8,325 images), and 15% test (8,325 images) sets for balanced evaluation
- Trained a Convolutional Neural Network (CNN) built from scratch in TensorFlow/Keras, consisting of three convolutional blocks (32–64–128 filters), dropout regularization (50%), and a dense classification head for multi-class output
- Implemented real-time inference via OpenCV with frame preprocessing (<120 ms/frame latency), class smoothing
TensorFlowKerasPythonOpenCVNumPyPandasMatplotlib
Priority Queue Data Structure (C++ SQueue Project)
Engineered a flexible, high-performance C++ priority queue (SQueue) supporting skew and leftist heap variants for advanced data handling.
Key Achievements:
- Implemented dual-heap architecture with O(log n) merge, insert, and extract operations
- Designed pluggable priority functions via function pointers for runtime flexibility
- Built a comprehensive test suite with random data generation and 100% pass rate
- Verified zero memory leaks using Valgrind and automated test coverage
C++MakefileValgrindData StructuresAlgorithms
Impact & Achievements
System Performance
ETF Optimizer
- • Analyzed 2,000+ ETFs across 30 categories using GPT-4o for portfolio optimization
- • Built real-time data pipeline for continuously updating ETF metrics from Yahoo Finance (API)
- • Deployed containerized backend on AWS EC2, ensuring smooth scaling under high API usage
Database Optimization
- • Processed 55,500 images for training a deep learning classifier using TensorFlow + Keras
- • Achieved high accuracy with optimized CNN architecture and data preprocessing pipeline
- • Enabled live translation from ASL to text with sub-120 ms inference latency
Infrastructure & DevOps
Cloud Deployment
- • Dockerized multiple projects and deployed to AWS EC2 with automatic build scripts
- • Implemented secure model and data storage in AWS S3
- • Reduced setup overhead by 40% through containerized deployment
Automation & Efficiency
- • Integrated multithreading in data preprocessing scripts to improve runtime
- • Built scalable backends using Flask and Firebase for multiple full-stack apps
- • Created CI workflows for reliable deployment and testing during fellowship projects
Development & Leadership
Software Engineering Fellow – Headstarter NYC
- • Collaborated on 5+ full-stack React/Next apps, integrating REST APIs and Firebase
- • Learned and applied 7 new frameworks and languages in a 3-month sprint environment
- • Delivered production-ready prototypes on strict deadlines within agile teams
Mentorship & Outreach – Verizon Robotics
- • Mentored underrepresented youth in software and robotics fundamentals
- • Enforced safety and curriculum standards while leading hands-on technical sessions
- • Promoted diversity and STEM engagement through guided engineering challenges
$ contact --info
Let's Connect
$ location --current
Rockville, MD
$ contact --email
ofc.tedsold@gmail.com$ my resume.pdf
Download Resume$ ls ./social-links
$ send-message