Featured Projects
- Built, deployed, and operate ktfo.gg, a live real-time MMA prediction platform with
a Go (Chi) backend, React/TypeScript frontend, PostgreSQL, Redis, and WebSocket-driven scoring,
results, and event state
- Architected a multi-tenant, event-driven backend using a transactional outbox, Redis Streams,
consumer groups, Redis Pub/Sub, and multi-room WebSocket routing for reliable low-latency
domain-event fan-out
- Reduced scoring pipeline latency from 14s → 2–3s under 10k simulated concurrent
users by replacing synchronous workflows with parallel batch processing and stream-based workers
- Designed a multi-dimensional scoring engine across winner, method, round, and time predictions, with
scorecard adjudication, crowd-percentage multipliers, and global/per-event Redis sorted-set
leaderboards
- Built server-authoritative live fight operations around a six-state finite state machine, PostgreSQL
CHECK constraints, optimistic concurrency control, clock correction tools, and a sequenced
play-by-play feed
- Shipped a shareable head-to-head analytics experience with composable user/event search, cumulative
and capture-rate charts, per-event totals, a sortable receipts table, and complete empty/error
states
-
Built a native SwiftUI/HealthKit app for analyzing real workout and heart-rate time-series data,
with deterministic metrics, interactive Swift Charts, recovery analysis, and data-quality checks
-
Designed a tool-calling analysis pipeline where LLMs can request deterministic workout analytics
and compose multi-section reports while numerical calculations remain fully application-controlled
-
Built a constrained generative-UI system using typed analytical artifacts, a runtime evidence
registry, dynamic JSON schemas, and native visual primitives, allowing models to choose supporting
visuals without generating arbitrary UI or invalid metric bindings
- Built an interactive React/TypeScript visualizer for A*, Dijkstra, BFS, and Greedy BFS with editable
weighted grids, heuristic controls, diagonal movement, and terrain-cost presets
- Designed snapshot and granular timeline playback with state-synchronized frontier, visited, path,
and cost overlays to expose each algorithm's decision process step by step
- Tested core search logic, heap behavior, path reconstruction, and f-cost calculations with Vitest
and GitHub Actions, including fallback-goal handling for unreachable destinations
Experience
- Developed an advanced data integration solution to unify multiple POI datasets, using string
similarity algorithms along with geographic distance to improve data accuracy and consistency
- Spearheaded a tenant finder recommendation engine utilizing the integrated POI dataset,
significantly enhancing matching accuracy and optimizing tenant placement
- Refined a large-scale corporate ownership dataset by conducting analytics, identifying and
correcting bad entries, implementing NER tagging, and consolidating duplicate data
- Engineered a robust ETL pipeline to extract data from a GraphQL API, implementing transformations
and normalization into seven tables across datasets ranging from 140M to 1B entries
- Implemented a comprehensive caching mechanism using Redis, enhancing data access speed and overall
system performance across the entire application
- Built a Redis-backed rate limiter with per-package keys and real-time usage tracking to optimize
resource allocation
- Contributed to the development of an enterprise API, focusing on scalability, security, and
integration capabilities
Education
First Class Honours · St. Augustine, Trinidad
Completed two years of coursework · St. Augustine, Trinidad
Skills
Languages
TypeScript, Go, Swift, SQL
Frontend
React, React Router v7, TailwindCSS, shadcn, Zod, dnd-kit, react-hook-form, neverthrow
iOS
SwiftUI, HealthKit, Swift Charts
Backend
Chi, REST, GraphQL, WebSocket, Event-Driven Architecture, Transactional Outbox, Redis Pub/Sub, Redis Streams
Databases
PostgreSQL, Redis, Snowflake, PostGIS, PgVector, BigQuery
Infra & Tools
Docker, GitHub Actions, CI/CD, Vitest