E-mail | GitHub | Scholar | CV

I want to beat Gurobi on equal machinery

My name is Francisco Galuppo and I like to solve optimization problems using neural nets. My Master's dissertation was about solving Multi-Stage Stochastic Optimization (MSSO) problems using tiny neural nets. The algorithm is called Stochastic Neural Dynamic Programming (SNDP). Now I want to use large nets, and for that I'll need a MIP solver faster than Gurobi's. That's what I've been doing recently.

Let's work together

I'm from Brazil btw. I hold Bachelor and Master degrees in Computer Science from the Federal University of Minas Gerais (UFMG). I wrote some papers. And I worked as a machine learning engineer for a while.

I write mostly in Python. I tried Julia but no one seems to document things there.

---

Passa-tempo

From time to time I like to implement pointless things:

Vehicle Path Planning: I simply like this problem very much. It is about achieving the most fuel-efficient route selection. The fundamental challenge lies in guiding the vehicles from their initial position to a set of targets, all while ensuring collision-free movement.

Mahjong Tracker: I like to play mahjong (Hong Kong Old-Style) with my friends, so I asked Claude Code to implement this tracker. It handles dealer and wind rotation, computes payouts based on fans and win type, and keeps a log of every hand.