| Desig.⇅ | Object⇅ | Mag.⇅ |
|---|
An extractive summarization framework with a web interface for scalable, real-time document summarization. Built with Python, spaCy, and Streamlit, and evaluated using ROUGE metrics.
Emission lines mark the technologies present in the toolchain — read like a stellar spectrum, each line is one instrument in regular use.
Built and deployed production ML models up to 87% accuracy, with REST APIs via FastAPI, Docker, and AWS cutting serving latency 35%.
Used structured patient data to predict breast cancer, diabetes, and hospital readmissions with SHAP interpretability for clinicians.
Built hybrid recommenders and NLP pipelines turning interaction and text data into targeted suggestions and content decisions.
Clarify metrics, success criteria, and constraints with stakeholders before touching a line of code.
Use Pandas, visualization, and profiling to understand distributions, leakage, and data-quality issues.
Train baselines first, then iterate with better features, architectures, and proper evaluation (CV, right metrics).
Turn results into dashboards and a simple story: what changed, by how much, and what to do next.
Strengthening probability, statistics, core ML algorithms, and clean analysis code in Python & SQL.
Building and deploying models for sequence data — stock prices, text, audio — with transformers and advanced architectures.
Moving from one-off notebooks to reliable systems: experiment tracking, versioning, monitoring, deployment.
A keyword-based approximation for demo only — real projects use trained models with proper evaluation.
Ask about my education, projects, experience, skills, or the roles I'm looking for. Runs fully in your browser — no external calls, no tracking.
Open to Data Scientist / Data Analyst / ML roles across the US. Recruiter, collaborator, or fellow sky-watcher — the channel's open.