[ 00 // STATUS: ONLINE ]

Vic
_

I build systems that move, think, and learn.

[ 01 // ABOUT ]

My journey began in pure computer science, but the end goal was always robotics. Before writing my first line of code, I knew I wanted to build machines that interact with the physical world. When deciding whether to approach this through hardware like mechanical engineering, electrical engineering, or software, my fascination with AI ultimately made the choice clear.

I chose the software path because of a specific vision: combining robotics with machine learning to create systems that feel genuinely "alive." Inspired by sci-fi like Iron Man, Big Hero 6, and Chappie, I realized that while pure hardware builds the body, AI builds the mind. Even computer engineering felt too focused on low-level constraints rather than pushing the boundaries of intelligent autonomy.

Currently, I'm focusing on machine learning and AI while continuously expanding my knowledge in classical robotics. My trajectory is laser-focused on merging these domains by using modern learned autonomy to bring complex physical systems to life.

→Building RL guidance systems.

→Optimizing reward functions for learned autonomy.

→Exploring Model Predictive Control.

FIG 1.0 — INTERDISCIPLINARY TRAJECTORY

[ 02 // PROJECTS ]

Featured

RL Missile Guidance System

GITHUB

Trained a PPO agent to intercept a moving, evasive target using single-axis thrust vector control — one continuous input, no throttle. The agent independently developed lead pursuit trajectories consistent with Proportional Navigation without being explicitly programmed with it. Trained across a 5-stage curriculum from static intercept to active evasion with obstacle luring.▶ The clip below shows the trained policy intercepting a human-controlled target in real time.

PythonPyTorchStable Baselines 3GymnasiumNumPyPyGamePPOCurriculum Learning

AI Gameplay Analysis System

GITHUB

End-to-end AI pipeline that analyzes raw gameplay footage using a vision-language model for perception and a dedicated reasoning model for pattern analysis. No game APIs — just video in, evidence-grounded analysis out.

PythonOllamaVLMLLMDocker

[ 03 // STACK ]

Languages

Python

Java

TypeScript

JavaScript

SQL

Rust

Frameworks & Libraries

PyTorch

Stable Baselines 3

ROS 2

Langchain/Langgraph

React/Next.js

OpenCV

Gymnasium

NumPy

Pandas

Transformers (Hugging Face)

Tools

Git

Docker

Ollama

Linux

Core Domains

Reinforcement Learning
Proximal Policy Optimization, Curriculum Learning, Reward Shaping, Sim-to-Real Transfer.

Classical Controls
PID, Model Predictive Control (MPC), Proportional Navigation.

Simulation & Modeling
Custom Gymnasium Environments, Rigid Body Physics, TVC Dynamics, Curriculum Design.

AI Systems & Pipelines
Vision-Language Models, Perception-Reasoning Separation, Prompt Engineering, Local LLM Deployment.

[ 04 // CONTACT ]

Let's build something.

Open to research roles, internships, and interesting problems.

Vic_