• Led design and development of products & technology for the company from pre-idea to more than USD9M in valuation.• Directed design and implementation of AI Empowered Interactive Games for Children, representing a 12x increase in company revenue, using Blender and Unity. • Spearheaded design and implemented scalable applications, AI & LLM inferencing solutions, serving thousands of concurrent daily users through using AWS EKS, Flutterflow, Flask, K8s, and other AWS cloud technologies.• Trained Speech and Phoneme Recognition Deep Learning Models to reach an above 92% 2-shot accuracy using local data in 3 months, enhancing productivity in English Lessons by 5x.• Developed and Deployed LLM Agents for Education shortening Lesson Planning Times by 60% on average using DPO, SFT, GraphRAG and Prompt Engineering.• Established a Global Development Team across 5 countries (Poland, Indonesia, Singapore, Japan, Hong Kong) through hosting and speaking in global events, conferences and networking sessions.

• Organised a global Tech for Good conference with 60 participants across more than 5 countries.• Designed (UI/ UX) and Built Full Stack Mobile Application for English Learning for Preschools, accelerating administrative productivity by 5x from 15 hours a week to 3 hours a week, using React and Express. • Architected and Implemented pipelines for the Deployment and Optimization of AI models to improve speeds of model inference by 5-20x and evaluation time by 10-25x.• Constructed pipelines for Continuous Integration/Deployment, pull requests, code reviews, load/stress testing, unit/integration/ system, end-to-end testing

• Designed and Built an Automated Surveillance and Communication Application reducing 24/7 surveillance manpower requirements by 16x, using Flask and Javascript. • Constructed an algorithm for Image classification using for high-resolution data (million by million pixels) feature extraction, improving processing speed by 10x with no loss in accuracy using OpenCV and SKLearn.• Designed an Optimization pipeline for all existing Production-grade DL models improving DL Inference speed by 2.3x, using Tensorflow and TensorRT.• Created Algorithms to Automatically Identify, Index and Log changes in Landmass with global coordinates with 95% accuracy, reducing operational manpower requirements by 20x, using DL and OpenGL.• Established a model training, evaluation, and deployment pipeline for object detection/ segmentation on local data with 90% mAP using Tensorflow and CVAT.