AdityaKatheeth

Model Card

An experiment in self-description, borrowing the format from Mitchell et al. (2019).


Intended Use	Applied ML engineering and algorithmic fairness research. Auditing high-stakes decision systems in credit, hiring, healthcare, and public-sector risk.
Out of Scope	Formal learning theory. I work on deployed systems and their real-world effects — not on mathematical foundations.
Training Data	M.S. in Artificial Intelligence. 5+ years in cloud data engineering and production ML. Self-directed study of the fairness and STS literature.
Evaluation Metrics	Production systems shipped · research notes published · open-source contributions · workshop and conference submissions (FAccT, AIES)
Limitations	Early in the transition from engineering to academic research. Actively building depth in causal inference, fairness theory, and AI governance.
Ethical Considerations	Algorithmic audits, disparate-impact measurement, model-card and datasheet documentation, governance frameworks for high-stakes deployments.

Engineering — production ML pipelines, feature stores, and model observability on modern cloud data platforms
Research — algorithmic audits, disparate-impact measurement, and documentation standards (model cards, datasheets) for deployed systems
Writing — distilling fairness research for practitioners; submissions targeted at FAccT and AIES

Python · SQL · PyTorch · scikit-learn · Spark · dbt · Terraform

$Weapons of Math Destruction — Cathy O'Neil$

_{Click any book for notes. Full index: ai-ethics-research-log/READING.md.}

Placeholders — real repo links land as projects ship.

fairness-audit-toolkit — reusable harness for running disparate-impact and equal-opportunity audits against tabular classifiers, with model-card generation
ai-ethics-research-log — public reading log and working notes for fairness and AI governance research
streaming-ml-platform — end-to-end Kafka → Spark Structured Streaming → feature store reference implementation

Research collaborations, audit work, and fairness consultations welcome.

_{GitHub · LinkedIn · Email}