I build AI products, intelligent systems, and data-driven solutions.

Financial valuation assumptions are frequently hidden in spreadsheet cells or opaque web models, making sensitivity reviews difficult before calculating enterprise and equity value.

Engineered a deterministic Python/FastAPI valuation engine decoupled from the UI, supporting dual Gordon Growth and Exit Multiple terminal value models with automated Excel (.xlsx) and JSON reconciliation.

Enforced strict Pydantic input schemas, mathematical determinism with zero cell-drift between UI state and exported workbooks, and isolated local-write checks.

High-intent customer inquiries often get lost or duplicated during third-party API spikes and external service connection dropouts.

Designed and built an event-driven lead ingestion pipeline implementing the transactional outbox pattern in PostgreSQL, with HMAC token authentication, automated deduplication, and Google Sheets webhook sync.

Enforced atomic database transactions storing events before outbound delivery, exponential backoff retries with bounded failure thresholds, and formula injection sanitization.

EV drivers struggle to quickly find operational, nearby charging stations along real travel routes when aggregators have stale data or slow radial search.

Engineered a full-stack station locator with PostGIS spatial indexing for sub-50ms proximity queries, bounding-box map view filtering, and verified rate-limiting boundaries.

Implemented PostGIS ST_DWithin spatial indexing on geographic coordinates, API payload limits, and debounce-optimized map bounds calculation.