Iโm a Data Scientist and Consultant with a multidisciplinary background. My journey spans research science, philosophy, teaching, and now business. This lets me strategically solve complex problems, primarily using multi-modal LLMs."๐ฆ๐ผ ๐๐ต๐ ๐ฑ๐ผ ๐๐ผ๐ ๐ต๐ฎ๐๐ฒ ๐ฐ ๐ฑ๐ฒ๐ด๐ฟ๐ฒ๐ฒ๐?"I started with Chemistry (with Maths) to be a research scientist, working on some of the world's largest experiments, like ๐๐๐ฅ๐ก in Geneva. But science and technology donโt exist in a vacuum โ they impact society. Thatโs why I pursued a Philosophy degree, to gain a deeper understanding of society, myself, and others.Iโm most motivated by helping people. I trained and worked in Education. It perfected my ability to explain complex topics, like technical requirements, to any audience. This skill is now central to my consulting work. I later moved into Data Science via business school. This taught me to ๐๐ต๐ถ๐ป๐ธ ๐๐๐ฟ๐ฎ๐๐ฒ๐ด๐ถ๐ฐ๐ฎ๐น๐น๐ about the big picture, so the systems I build also align with companies' long-term data strategies and business goals.I'm grateful to have had enough experience to be able to collaborate and communicate well. I have solid experience managing both myself and teams, leading technical and non-technical teams smoothly โ adults are easier than remotely teaching maths to unruly teenagers during lockdown!I strongly believe those who effectively integrate AI will outperform those who donโt. That conviction drove me to Nigeria in 2024 to present on AI, and I continue to mentor students I met there.Thankfully, my technical ML experience is up to date: my specialty is generative AI and LLMs (RAG, data extraction, zero-shot classification, generative pipelines etc.). I love attending Python conferences, and you'll typically find me using PyTorch, XGBoost, Azure, HuggingFace, web scraping, and interactive Plotly data dashboards.Notable Projects:๐ ๐ฆ๐ฐ ๐ง๐ต๐ฒ๐๐ถ๐: Used multi-modal LLMs and web scraping to auto-generate investment reports for a VC. Delivered a pipeline that that was 50% less than the expected budget.๐ฃ๐๐๐ผ๐ป ๐๐ถ๐๐ต๐๐ฎ๐ป๐ถ๐ฎ ๐ง๐ฎ๐น๐ธ: Presented to 300+ attendees on ensuring LLM output consistency with clean OOP Python code and correctly predicted OpenAI's API updates 6 months in advance.๐๐ผ๐บ๐ฝ๐๐๐ฒ๐ฟ ๐ฉ๐ถ๐๐ถ๐ผ๐ป ๐ณ๐ผ๐ฟ ๐ง๐ฒ๐ป๐ป๐ถ๐: Evaluated competing computer vision models to automatically generate tennis video highlights, extracting only the 40% of tennis videos containing actual action.Iโm open to collaborations and eager to help unlock AIโs full potential โ if that resonates, letโs connect.
