I like understanding systems end to end. I'm currently exploring frontend design, git workflows, backend development, full-stack development, physical simulations, and project management β building hands-on in each rather than just reading about them. Where I've gone deepest is automation: I design and ship n8n + WhatsApp API workflows that qualify leads and save businesses real time and money, which is the core of what I'm building at Adesse.
Founder & Developer β Adesse Building automation systems with n8n and the WhatsApp API that qualify leads and save businesses time and money.
Web Developer β VantTec VantTec is a multidisciplinary student team at TecnolΓ³gico de Monterrey that designs, builds, and programs unmanned autonomous vehicles. I build and maintain the team's official website (React).
TecnolΓ³gico de Monterrey β B.S. in Computer Science High-performing student β GPA 99/100
Hermes AI-powered news pipeline: collects stories from NewsAPI and RSS, curates and summarizes them with the Claude API, and delivers a daily digest straight to Telegram.
CSV-Analyzer
Interactive C++20 CSV analysis tool with adaptive type detection, template metaprogramming (std::variant), and statistical summaries rendered as ASCII histograms. Final project for Object-Oriented Programming.
Electromagnetic Particle Confinement Β· live demo Simulated a charged particle confined by magnetic fields β from the Lorentz force to a 3D tokamak reactor. Implemented the RK4 integrator and the active, sectorized control system that keeps the particle confined in real time.
π HackMTY β Smart Intelligence β 1st place Airline catering analytics platform (FastAPI, Snowflake, Random Forest, Gemini AI). My part: API integrations and the frontend. Team project β full credit to the team in the repo.
π Capital One Sprint β MatchUp β 1st place Pick-up sports matchup app (Next.js, Supabase). My part: backend and the Supabase database. Team project β full credit to the team in the repo.
π± My GitHub Data
π¦ 262.4 kB Used in GitHub's Storage
π 398 Contributions in the Year 2026
π« Not Opted to Hire
π 12 Public Repositories
π 9 Private Repositories
I'm an Early π€
π Morning 779 commits βββββββββββββββββββββββββ 46.56 %
π Daytime 515 commits βββββββββββββββββββββββββ 30.78 %
π Evening 301 commits βββββββββββββββββββββββββ 17.99 %
π Night 78 commits βββββββββββββββββββββββββ 04.66 %
π I'm Most Productive on Monday
Monday 370 commits βββββββββββββββββββββββββ 22.12 %
Tuesday 284 commits βββββββββββββββββββββββββ 16.98 %
Wednesday 159 commits βββββββββββββββββββββββββ 09.50 %
Thursday 204 commits βββββββββββββββββββββββββ 12.19 %
Friday 298 commits βββββββββββββββββββββββββ 17.81 %
Saturday 268 commits βββββββββββββββββββββββββ 16.02 %
Sunday 90 commits βββββββββββββββββββββββββ 05.38 %
π This Week I Spent My Time On
ποΈ Time Zone: America/Bogota
π¬ Programming Languages:
C++ 3 hrs 51 mins βββββββββββββββββββββββββ 38.72 %
Assembly 2 hrs 23 mins βββββββββββββββββββββββββ 24.07 %
Python 2 hrs 1 min βββββββββββββββββββββββββ 20.41 %
HTTP 42 mins βββββββββββββββββββββββββ 07.10 %
C 39 mins βββββββββββββββββββββββββ 06.64 %
π₯ Editors:
VS Code 8 hrs 32 mins βββββββββββββββββββββββββ 85.78 %
Copilot CLI 58 mins βββββββββββββββββββββββββ 09.74 %
Claude Code 26 mins βββββββββββββββββββββββββ 04.48 %
π±βπ» Projects:
VS-Code 4 hrs 31 mins βββββββββββββββββββββββββ 45.41 %
Ensamblador 2 hrs 23 mins βββββββββββββββββββββββββ 23.96 %
HW - E04 1 hr 32 mins βββββββββββββββββββββββββ 15.48 %
API_Sensores_DiegoGallo 39 mins βββββββββββββββββββββββββ 06.66 %
Captura_Imagenes_ESP 38 mins βββββββββββββββββββββββββ 06.50 %
π» Operating System:
Windows 9 hrs 57 mins βββββββββββββββββββββββββ 100.00 %
π€ AI Coding This Week
β± AI Coding Time: 2 hrs 49 mins (28.43%)
βοΈ 1,504 lines written by AI, 573 lines written by hand (72.41% AI-written)
π€ 427,230 Input Tokens, 78,645 Output Tokens
π΅ $2.85 Estimated AI Cost This Week
π§ 7 AI Sessions, 46 AI Prompts
Opus 791 lines βββββββββββββββββββββββββ 52.21 %
Sonnet 655 lines βββββββββββββββββββββββββ 43.23 %
GPT 69 lines βββββββββββββββββββββββββ 04.55 %
π AI Coding Insights:
π€ AI-Driven β 72.41% of written lines came from AI
π Detailed Prompter β average 589 characters per prompt
π Iterative Prompter β average 7 prompts per session
π High AI Trust β 31.32% of changed lines were hand-edited
Timeline
Last Updated on 24/10/2026 21:08:23 UTC
Beyond code: philosophy, functional fitness, discretionary trading, and documenting the process.



