Work
Sizzle
Sizzle
Generative AI Engineer
June 2024 - Present
Actualizing the full pipeline for data processing, model training, and deployment of nutritional factors and macro nutrients for any given recipe.
- Integrated AI solutions into Google Cloud services using VertexAI to streamline operations.
- Improved data analysis capabilities from custom REST API endpoints, resulting in an easily scalable solution.
Upwork, Fiverr
Upwork, Fiverr
AI/ML Freelancer
August 2022 - Present
Developing and deploying custom AI/ML models for various clients.
- Improved business processes and outcomes using both open-source and paid LLMs, resulting in an average efficiency increase of 25%.
- Engineered end-to-end MLOps solutions, including CI/CD pipelines, model monitoring, and infrastructure automation, attaining a 98% customer satisfaction rate.
- Collaborated with cross-functional teams to integrate AI solutions into existing systems, enhancing overall efficiency and functionality, reducing operational costs by up to 20%.
Joint Genome Institute
Joint Genome Institute
Bio-Informatics Data Analyst
May 2020 - August 2021
Utilized R with a cross-functional team of researchers to construct workflows for genetic homology comparison.
- Slashed processing time by 90%.
- Constructed a model based on plant species datasets for microbial interactions, achieving 98% accuracy.
- Created and optimized genome reference pipelines, decreasing processing time from 4 hours to 10 minutes.
- Showcased and presented technologies to over 1000 end clients and senior leadership members in yearly user meeting.
Vice Lab
Vice Lab
Research Scientist
January 2020 - May 2021
Devised and performed Bash scripts to run hydro-power models on a proprietary supercomputer.
- Achieved a 50% reduction in processing time through asynchronous execution deployed with Slurm.
- Drove creation of a data visualization web application for hydro power models operated with Vue.js, increasing user engagement by 25% and improving data interpretation.
- Optimized model implementation and resolved bottlenecks using Cython, further cutting down execution time by 20%.