Data PlatformsFinancial Data Platform
End-to-end banking analytics platform that turns FDIC bank failures, FRED/ALFRED macroeconomic series, QBP/NIC institution data, and SEC filing context into a reproducible financial intelligence stack. The project demonstrates raw landing zones, warehouse-ready modeling, quality enforcement, orchestration, API/dashboard serving, and documentation surfaces designed for regulatory-style analytical review.
Engineering signal: Connector readiness checks, source traceability, dbt-style transformations, Great Expectations validation, Airflow orchestration, DuckDB/Snowflake-ready analytical layers, FastAPI health endpoints, Streamlit dashboards, Dockerized local execution, Terraform scaffolding, and CI-oriented project structure.
Data PlatformsPrivacy Engineering & Corpus Intelligence
Local-first corpus intelligence pipeline for classifying large exported conversation archives into public-safe, private, and review-required outputs. It streams split JSON exports, recovers safe chunks from mixed conversations, detects hard identifiers, routes sensitive domains, scores public topic families, and produces Markdown/JSON review artifacts without requiring bulk hosted-model inference over private text.
Engineering signal: Deterministic PII detection, Presidio/spaCy validation, semantic scoring, policy-rule gates, ensemble agreement checks, review queues, reproducible artifact generation, and privacy-preserving data preparation suitable for documentation, search, or knowledge-base publishing workflows.
Data PlatformsSecure Data Ingestion
Authentication-event ingestion platform that processes login telemetry through a reproducible local pipeline, validates the event contract, pseudonymizes sensitive identifiers, persists curated records, and publishes compact evidence views instead of exposing full source datasets. The system is built to show secure ingestion behavior from raw event intake through audit-ready outputs.
Engineering signal: SQS-compatible intake semantics, versioned event contracts, HMAC tokenization, deterministic event IDs, PostgreSQL landing and curated tables, quarantine paths, batch-level audit evidence, Docker Compose/LocalStack runtime, FastAPI preview endpoints, structured logging, and validation tests.
Data PlatformsConnected-Vehicle Telemetry
Connected-vehicle telemetry lakehouse that unifies live OBD/mobile payloads, profile-aware simulator events, and file replay into one session-aware contract. It models raw and normalized telemetry, supports operational dashboard APIs, exposes stream/export paths, and treats simulator, relay, replay, dashboard, and downstream consumers as clients of one trusted vehicle event model.
Engineering signal: Canonical event contracts, raw/normalized lakehouse tables, SQLite/PostgreSQL storage modes, live REST intake, session APIs, simulator controls, SSE dashboard streaming, Bronze/Silver/Gold transform paths, schema/export endpoints, and signed or encrypted exchange patterns for future services.
Data PlatformsPersonal Health Lakehouse
Privacy-first local health data platform for ingesting Apple Health export ZIP baselines and ongoing Health Auto Export JSON updates into a PostgreSQL medallion warehouse. The platform preserves raw evidence, deduplicates by file and payload hash, tracks lineage and quality events, and produces BI/agent-facing views without moving sensitive health data into hosted analytics systems.
Engineering signal: Immutable archive flow, REST and folder-drop intake, stable-file waits, SHA-256 dedupe, health_meta lineage tables, bronze raw preservation, silver conformance, gold temporal rollups, health_mart views, FastAPI receiver, worker service, dbt transform scaffolding, Grafana/Prometheus/Loki monitoring, and parser/hash/API tests.