Data Science · Data Analytics · Machine Learning · Software Engineering
Physics PhD researcher building reproducible data, ML, and software systems — with a focus on analytics workflows, model evaluation, backend-supported experiment platforms, and scientific software.
I focus on turning messy, ambiguous, or computation-heavy problems into structured, testable, and reproducible systems.
- Data Analytics & Data Science: data quality, SQL marts, feature engineering, time-aware validation, statistical diagnostics.
- Machine Learning: controlled experiments, baselines, multi-seed evaluation, calibration, error analysis, model serving.
- Backend & Experiment Platforms: FastAPI services, database-backed run state, async workers, Docker/CI reviewer paths.
- Scientific & Numerical Software: Julia packages, external model-fitting workflows, residual diagnostics, parameter search, ODE-based simulation.
- For Data Analytics / Data Science: Wearable Analytics — SQL marts, data quality, feature engineering, time-aware validation.
- For Machine Learning / ML Engineering: PyTorch Pets Classifier — MLflow experiments, multi-seed evaluation, diagnostics, FastAPI/Docker deployment.
- For Backend / Experiment Platforms: Evalynx — FastAPI, PostgreSQL, Redis/RQ, async run orchestration, CI smoke tests.
|
Privacy-first analytics workflow on real Garmin data: sanitized pipelines, quality labels, SQL marts, feature engineering, leakage-aware modeling, and tests. Signals: Data Analytics · Time Series · SQL · ML Evaluation |
Reproducible computer-vision workflow with MLflow tracking, multi-seed model selection, diagnostics, calibration, FastAPI serving, Docker, and Azure deployment path. Signals: Machine Learning · Model Reliability · Deployment |
|
FastAPI/PostgreSQL/Redis control plane for reproducible computational runs: API submission, async workers, retries, metrics, artifacts, and CI smoke tests. Signals: Backend · Run Orchestration · Experiment Platform |
Deterministic hidden-information simulator with legal-action masks, replay traces, fixed-seed benchmarks, heuristic/search/learned agents, and evaluation tooling. Signals: Simulation · AI Agents · ML Evaluation |
|
Julia framework for external model-fitting workflows with structured metrics, residual diagnostics, white-noise fitting, priors, and adaptive parameter search. Signals: Statistical Diagnostics · Experiment Orchestration · Scientific Computing |
Julia scientific-computing package for ODE-based simulation workflows with equation-of-state abstractions, shooting methods, parameter scans, and sensitivity analysis. Signals: Numerical Methods · Sensitivity Analysis · Julia |
- documented setup and reproducible run paths;
- tests, CI checks, and structured artifacts;
- baseline comparisons, metrics, diagnostics, and failure cases;
- clear separation between data processing, modeling, evaluation, and serving;
- public repositories with code, documentation, and project-specific README files.
Across projects, I emphasize:
- explicit data contracts and clear interfaces;
- reproducible runs and inspectable intermediate artifacts;
- baselines, diagnostics, and conservative validation;
- automated tests, documentation, CI, and reviewer-friendly execution paths;
- translating complex domain problems into maintainable software.
Open to roles in Data Science, Data Analytics, Machine Learning Engineering, and Software Engineering where reproducibility, evaluation, and analytical systems matter.
