Machine Learning Engineer with expertise in computer vision, deep learning, and reinforcement learning. Currently working at Smartlens Inc. supporting 110+ FDA clinical studies with ML model integration. Cornell University Computer Science student with a strong background in AI research and practical ML applications.

• 🔭 Currently working as ML Engineer at Smartlens Inc.
• 🌱 Specializing in Computer Vision, Deep Learning, and Reinforcement Learning
• 🎓 B.Sc. in Computer Science, Minor in AI & ORIE (Cornell University, May 2025)
• 📫 Contact: alpkomban@gmail.com
• 📍 Based in Palo Alto, CA
• 🎯 GPA: 3.8/4.0

• Engineered features for image regressor achieving 85% accuracy with 10ms inference speed for video frame quality evaluation
• Supporting 110+ FDA clinical studies by integrating 5 ML models (2 classification, 2 segmentation, 1 regressor) into evaluation pipeline
• Built Vision Transformer model improving lens level identification precision to ±1 (5x improvement over previous CNN model)

• Built neural network models to identify optimal frames from 60+ minutes of video data, reducing analysis time to 1 minute per dataset
• Developed segmentation algorithms achieving 70% accuracy in extracting key anatomical components from medical frames
• Analyzed performance of 3 models across 3,000+ images and built CNN regressor with ±5 precision for lens level assessment

• Developed e-commerce website managing 300+ products and processing 600+ orders/month
• Built responsive UI components for product catalogs and administrative dashboards supporting 300+ SKUs
• Helped transition business to independent platform, reducing platform fees

• Developed computer vision application with Swift Engineers utilizing multiple CNN architectures for medical imaging analysis
• Implemented image classification and segmentation models with high precision to identify and isolate key anatomical structures
• Built image regression pipeline combining CNNs and Vision Transformers (ViTs) for quantitative assessment

• Developed encoder-decoder network to embed hidden messages into images while maintaining visual quality
• Implemented Reed-Solomon Bits Per Pixel (RS-BPP) evaluation metric to measure steganographic capacity
• Achieved data hiding capacity of 4.4 RS-BPP with maintained image realism and 0.59 auROC detection avoidance

• Developed custom Gymnasium environment and trained multiple RL policies for simulated bipedal robot walking using MuJoCo
• Implemented and compared three reinforcement learning algorithms: PPO, SAC, and DDPG
• Designed neural network architectures and reward functions for standing and walking tasks with extensive hyperparameter tuning

• Machine Learning: Supervised/Unsupervised Learning, Reinforcement Learning, Computer Vision
• Deep Learning: CNNs, Vision Transformers (ViTs), GANs, LLMs, Encoder-Decoder Networks
• Computer Vision: Image Classification, Segmentation, Object Detection, Medical Imaging
• Reinforcement Learning: PPO, SAC, DDPG, Custom Gymnasium environments
• Optimization: Stochastic Processes, MDPs, Mathematical Modeling

Cornell University, College of Engineering | Ithaca, NY
Bachelor of Science in Computer Science, Minor in AI & ORIE | May 2025
GPA: 3.8/4.0

Relevant Coursework: Deep Learning, Computer Vision, Natural Language Processing, Foundations of Artificial Intelligence, Introduction to Machine Learning, Database Systems, Analysis of Algorithms, Optimization I & II, Stochastic Processes for Decision-Making

• Supporting 110+ FDA clinical studies with ML model integration
• Achieved 5x improvement in lens level identification precision with Vision Transformers
• Built models handling 60+ minutes of video data with 1-minute analysis time
• Processed 3,000+ medical images for model training and evaluation
• Helped over 600 students master programming concepts as Teaching Assistant