Alfie Pelicano

Research Specialist & Geodetic Software Developer

Where tectonic data meets tractable software

"All models are wrong, but some are useful." — George E. P. Box

I work at the intersection of earth science and software engineering — processing GNSS and geophysical data for earthquake research and fault monitoring, while building the software to automate what shouldn’t still be manual. With 16+ years in scientific computing and spatial data, I design headless processing pipelines, distributed task systems, and full-stack applications. I prefer minimalist, low-cognitive-load architecture: build to make impossible or tedious work tractable, not to flex a complex stack.

GitHub View Projects

Featured Projects

Earth science automation, civic platforms, and algorithmic tools — built to make hard things tractable

🌋 Earth Science & Field Automation

🛰️
Active Development

MOVE Faults

Flagship GNSS automation monorepo replacing decades of fragmented legacy scripts with a modern distributed processing stack.

Problem: Decades of fragmented MS-DOS/Bash GNSS processing scripts with no orchestration, no fault tolerance, and no field-usable interface.
Solution: Celery-backed distributed task queue, real-time NMEA TCP streaming, and an offline-capable React field app for live fault monitoring operations.
Stage: Active development — core task queue and streaming pipeline operational; field app in integration testing.
Python Celery Redis React FastAPI Docker
View Code
🌋
Active Development

EIL Hazard Suite

Decoupled geohazard system: a headless spatial processing engine for landslide susceptibility paired with a React-based ground-truth validator.

Problem: Geohazard susceptibility analysis required manual, error-prone GIS desktop workflows with no programmatic access or field validation tooling.
Solution: EIL-Calc: headless spatial processing engine for landslide susceptibility scoring. EIL-Viz: React visualizer for overlaying and validating results against ground-truth field data.
Stage: Active development — EIL-Calc processing pipeline stable; EIL-Viz in active integration.
Python FastAPI React Spatial Analysis
View Code
🗺️
In Production

PHAST

Chrome Extension command center that injects professional GIS controls into Leaflet maps and auto-generates official PHIVOLCS hazard assessment text.

Problem: PHIVOLCS hazard assessors had no professional GIS controls in standard web mapping tools, requiring manual text authoring for every official assessment output.
Solution: A Chrome Extension that injects GIS measurement and markup controls directly into Leaflet maps, then dynamically generates formatted official hazard text from map interactions.
Stage: Active development — core GIS injection and text generation functional; expanding to additional assessment templates.
Vanilla JS Chrome Extension API Leaflet.js
View Code
📋
Active Development

SIPAT

Open-source CAPI alternative to enterprise survey tools — built for field researchers who need offline-first data collection with complex survey logic.

Problem: Enterprise CAPI tools like SurveyToGo are expensive, closed-source, and impractical for field research teams operating in low-connectivity environments.
Solution: Offline-first data collection with complex branching survey logic, automated audio transcription workflows, and open data export — designed for field researchers, not enterprise procurement.
Stage: Active development — core offline collection and survey logic complete; transcription pipeline in integration.
TypeScript React IndexedDB Web Audio API
View Code
📂
Active Development

Drive Archaeologist

Python CLI for excavating and organizing legacy GNSS (GPS) data from old drives with inconsistent file structures.

Problem: Decades of GNSS data scattered across drives: random directories, duplicates, and missing metadata — making retrieval and reuse difficult.
Solution: Automated scanner that detects GNSS file types (RINEX and others), extracts metadata, flags duplicates, recommends reorganization, and generates safe migration scripts.
Stage: Phase 1/4 — foundation scanning and classification; built for geodesists managing legacy scientific archives.
Python 3.11+ Click CLI Rich Terminal Pandas Pytest
View Code

💻 Full-Stack Platforms & Civic Tech

🏥
Active Development

YAKAP-Link

Civic tech proof-of-concept addressing supply chain and dispensing bottlenecks in the PhilHealth YAKAP medicine program.

Problem: The PhilHealth YAKAP program suffered from phantom claims and medicine shortages due to untracked dispensing and no transaction guard logic.
Solution: Strict transaction constraints and dispensing guard logic to mitigate phantom claims, enforce availability checks, and ensure medicine reaches intended beneficiaries.
Stage: Proof-of-concept — core constraint and dispensing logic implemented; pending stakeholder review for pilot deployment.
Next.js TypeScript PostgreSQL Prisma
View Code
🔬
In Production

PipetGo

B2B laboratory marketplace MVP connecting businesses with ISO-accredited testing labs, with multi-tiered pricing and role-based authorization workflows.

Problem: Businesses had no centralized source for accredited labs — data was scattered, outdated, and often missing accreditation details.
Solution: A searchable marketplace with three pricing modes (QUOTE_REQUIRED, FIXED, HYBRID), multi-role authorization, and RFQ workflows connecting certified labs and clients directly.
Stage: MVP undergoing User Acceptance Testing with initial partner organizations before pilot deployment.
Next.js 14 TypeScript Prisma NextAuth Zod Vitest
View Code
🏠
Active Development

Homebase

Multi-residence household command center built for ADHD executive function support — aggressive, context-aware reminders that catch what the brain won't.

Problem: Passive reminder apps fail ADHD brains — notifications are dismissible, don't understand context (which residence you're at), and never escalate urgency. The result: forgotten utility bills and missed maintenance with hard-to-reverse downstream consequences.
Solution: ADHD-optimized urgency system (color-coded, impossible to ignore) with multi-residence bill tracking, recurring obligation engine, heuristics for anomaly detection and smart suggestions, chores management, and a distributed notification loop via daily cron with SMS fallback planned.
Stage: Phases 1 & 2 complete — deployed on Vercel, 32 E2E tests passing, alpha testing with 5 family users. Distributed hexagonal architecture implemented for portability to self-hosted hardware (Phase 12+).
Next.js 14 TypeScript Drizzle ORM Neon (PostgreSQL) Vercel Cron Playwright
🚙
Active Development

Carpool Coordinator

A community carpooling platform that helps residents post and coordinate ride shares with live updates and vehicle info.

Problem: Carpool plans were managed in chat threads — easy to lose track of, and difficult to match riders by time or route.
Solution: A structured posting system with schedule matching, vehicle details, and interest tracking for organized community rides.
Impact: Deployed on Render for a residential group; improved participation rates and reduced coordination time for parents and commuters.
Node.js Express PostgreSQL Vanilla JS Playwright
View Code
Washboard dashboard showing real-time queue management with customer bookings
🚿
Active Development

Washboard

Car wash queue and booking system that replaces pen-and-paper operations with contactless QR bookings and a real-time dashboard.

Problem: Paper-based queue tracking offered zero visibility, required manual position management, and caused long waits during peak hours.
Solution: Contactless QR bookings combined with a receptionist dashboard that provides live queue updates, position reordering, magic-link check-ins, and status tracking.
Impact: Production-ready (130/130 tests passing, 94/100 security score); piloting with car wash operators to refine real-world workflows.
Next.js 14 React 19 TypeScript PostgreSQL Passport.js Vitest
View Code

🧮 Algorithms, Architecture & Utility

⚡
Active Development

UGT (Chaotic Circuits)

Modern Python/LaTeX reconstruction of a 2004 physics undergraduate thesis — numerical solvers for Lyapunov Exponents in chaotic Sprott circuits.

Problem: Original 2004 thesis existed only in legacy formats with no reproducible computational environment or modern numerical tooling.
Solution: Full reconstruction with modern Python numerical solvers computing Lyapunov Exponents, phase portraits, and bifurcation diagrams for chaotic Sprott circuit systems. Published as reproducible LaTeX document.
Stage: Complete — solvers and LaTeX output published; serves as a reference implementation for chaos theory in physical circuits.
Python NumPy SciPy Matplotlib LaTeX
View Code
➕
In Production

Window Cards Generator

Zero-dependency Vanilla JS constraint-solving engine that dynamically generates printable math problem grids — designed for classrooms with limited device access.

Problem: Teachers needed customizable math materials without recurring costs from commercial workbook sets, usable on any device with no installation.
Solution: A browser-based constraint solver that generates arithmetic worksheets with strict operational constraints on demand — printable, adjustable by range and operation type, with zero dependencies.
Impact: Used in SPED and early-grade classrooms — providing a free, flexible alternative to paid materials with no device or connectivity requirements.
Vanilla JS HTML5 CSS3 Print CSS
Live Demo View Code
🎨
In Production

ArtPortfolio

High-performance statically generated Astro CMS built over a weekend — optimized for minimal cognitive load and fast asset delivery for digital art commissions.

Problem: Existing digital art commission sites were slow, cognitively noisy, or required ongoing CMS subscriptions — not practical for a solo artist portfolio.
Solution: Statically generated Astro site with zero client-side JS by default, optimized image delivery, and a content structure designed to minimize decision fatigue for commissioners.
Impact: In production as a live art commission site — fast, cost-free to host, and maintainable without a CMS subscription.
Astro Static Site Generation GitHub Pages
View Code
🤖
Active Development

NightCoder Orchestrator

Python-based automation engine that reads ADRs, generates structured implementation plans, and produces git-style patches using local LLMs (via Ollama).

Problem: Engineering tasks tied to ADRs required repetitive planning and manual patch authoring across multiple repos — slow, error-prone, and inconsistent.
Solution: A multi-agent runner that scans ADRs, synthesizes a step-by-step plan, generates patches, logs runs, and optionally applies changes through a safety-gated workflow.
Stage: Phase 2/4 — stable planner/coder loop; refining patch validation and multi-repo robustness.
Python 3.11+ Ollama deepseek-r1 (1.5B) qwen2.5-coder (1.5B) Subprocess / Shell Scripts Markdown ADR Parsing
📄
Active Development

SciWriter Orchestrator

Automated writing pipeline that converts research notes into structured drafts — with refinement passes, terminology checks, LaTeX export, and optional slide generation.

Problem: Drafting scientific manuscripts from scratch is slow, inconsistent, and prone to terminology drift across revisions.
Solution: A configurable multi-stage workflow that generates a draft, performs iterative refinement, normalizes terminology, and outputs LaTeX or slide formats.
Stage: Phase 2/4 — pipeline functional with draft/refine/terminology steps; improving LaTeX and slide output consistency.
Python 3.11+ Ollama qwen2.5 (3B Instruct) Shell Script Pipeline LaTeX / Beamer Config-Driven Workflow

Technical Skills

Frontend

  • Next.js 14 (App Router)
  • React 19 (Server Components)
  • Astro (Static Site Generation)
  • React / Vite
  • TypeScript
  • Vanilla JavaScript
  • Tailwind CSS
  • shadcn/ui

Backend

  • Python / FastAPI
  • Node.js / Express
  • Next.js API Routes
  • Celery (Distributed Task Queues)
  • Redis
  • Docker
  • Agentic AI Workflows (LangGraph / Ollama)
  • Prisma ORM
  • NextAuth.js / Passport.js
  • Zod Validation

Database & Infrastructure

  • PostgreSQL (Advanced SQL)
  • Neon (Serverless)
  • Supabase (PostgreSQL)

Python & Automation

  • Python 3.11+
  • Click CLI / Rich Terminal
  • Pandas (Data Processing)
  • NumPy / SciPy
  • Pytest (Testing)
  • Shell Scripting

Domain Expertise

  • GNSS Processing & Field Ops
  • Spatial Geohazards (Landslide / Seismic)
  • Heuristic Engines & Constraint Solving
  • Scientific Computing & HPC
  • Fortran (Legacy Scientific Code)

Testing & Quality

  • Playwright (E2E)
  • Vitest (Unit)
  • Pytest (Python)
  • Git Workflows

Tools & Workflows

  • Agentic Workflow Design & Orchestration
  • LaTeX / Beamer
  • ADR-driven Architecture
  • Linux / Bash

About Me

I work at PHIVOLCS processing GNSS and geophysical data for earthquake research and fault monitoring — and I build the software to automate what shouldn’t still be manual. The two roles aren’t separate: 16+ years of scientific computing taught me that precision, edge-case handling, and long-term maintainability aren’t engineering virtues, they’re survival requirements when your data tracks how the ground moves.

My philosophy is simple: build things that solve real problems for real people. I don’t build for the sake of flexing a complex stack — I build to make impossible or tedious work tractable. No moonshot hype, no blockchain detours.

I believe in:

  • Test-driven development — if it’s not tested, it’s not done
  • Systematic problem-solving — big problems get smaller when you break them right
  • Clear documentation — your future self deserves clarity
  • Practical design — tech should help people, not just impress developers

When I’m not writing code or processing fault data, I’m helping my kids with homework or automating something that probably didn’t need automating — because curiosity doesn’t switch off after work.

Let's Connect

Building something interesting? Need a developer who actually ships?
Let's talk.

alfieprojects.dev@gmail.com GitHub