Machine Learning Developer

SKDIV® Inc, Hyderabad, India

• Engineered and deployed hybrid recommendation algorithms (collaborative filtering, content-based, and ensemble methods) for CREW™, a consumer product platform, improving recommendation relevance by 18% and driving a 12% increase in user engagement
• Built end-to-end data pipelines to ingest, clean, and pre-process multi-source data at scale, ensuring consistency and integrity for downstream model training on 200K+ user interaction records
• Designed evaluation frameworks (precision@k, NDCG, CTR, engagement and satisfaction metrics) to benchmark recommendation quality and guide iterative model improvements
• Optimized model architecture and inference pipeline for scalability and low-latency real-time serving, reducing inference latency by 25% and supporting 500+ requests per second
• Partnered cross-functionally with product and engineering teams through design reviews and iterative feedback cycles, aligning model behavior with product goals and user needs

Full Stack Developer

Highbrow Technology Inc, Cary, United States

• Refined Al Japanese language model in collaboration with the Turing engineering team, boosting model consistency by 20% and improving reliability for production use
• Designed and tested prompts for Anthropic Cloud Al Japanese language model, increasing model accuracy by 15% across real-world datasets
• Analyzed 120,000+ model outputs, identified inconsistencies, and provided actionable feedback that reduced errors by 12% and improved overall performance
• Collaborated with Al researchers, engineers, and QA teams to optimize model training and validation pipelines, ensuring faster and more accurate iterations
• Built and maintained 3 internal full-stack tools to automate workflows for Al researchers and engineers, saving 15 hours of manual work per week

Full Stack Developer

Turing, Palo Alto, United States

• Refined a Japanese-language LLM (Anthropic Claude) in collaboration with Turing's AI engineering team, improving model consistency by 20% and accuracy by 15%
• Analyzed 120,000+ model outputs, identifying inconsistencies and delivering structured feedback that reduced errors by 12%
• Built 3 internal full-stack tools to automate researcher/engineer workflows, saving 15 hours/week of manual work
• Collaborated cross-functionally with AI researchers, engineers, and QA to optimize training and validation pipelines

Blockchain developer

Edmonton Global, Edmonton

• Led the development of a decentralized application (dApp) that utilizes AI for predictive analytics in trading, integrating machine learning models with blockchain for secure transaction verification.
• Implemented AI-driven algorithms that analyze market trends and predict price movements, increasing the accuracy of trades by 40%.
• Developed smart contracts on Ethereum to automate trading processes, enhancing security and transparency for users.
• Used Hardhat and Truffle to develop upgradeable Solidity contracts, ensuring flexibility and a streamlined deployment process.
• Created a Solidity escrow contract with external adapter integration to securely release funds based on specific outcomes, ensuring reliable marketplace transactions.

Fullstack developer

TakeMe, Tokyo

• Developed an advanced DNN-VAD system for a trap shooting application, employing Python and VAD libraries for real-time voice activity detection.
• Achieved a 30% increase in speech recognition accuracy through CNN model training with TensorFlow and Keras, incorporating data augmentation techniques to enhance model robustness.
• Increased sales by 45% and improved customer satisfaction by 60% through web development and optimization using Next.js and SCSS.
• Implemented strategies that improved performance by 30%, with a focus on optimizing product search speed.
• Engineered a backend system with PHP, Node.js, and Python, reducing server response time by 40% and improving performance for over 100,000 users.

Web developer Intern

Nulab Inc, Fukuoka

• Contributed to a project where I helped create a responsive e-commerce website, utilizing HTML, CSS, and JavaScript to enhance user experience.
• Collaborated with the team to integrate payment processing and inventory management features, gaining valuable experience in both front-end and back-end development.

Master of Engineering

Nagoya Institute of Technology

Bachelor of Engineering

Nagoya Institute of Technology