Skip to content
View Thom-320's full-sized avatar

Highlights

  • Pro

Block or report Thom-320

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Thom-320/README.md

Thomas Chisica Londoño

Applied Mathematics & Computer Science · Universidad del Rosario · Bogotá, Colombia

CV · Email · Project guide

What I work on

I build simulations and learning experiments, then test what their comparisons actually establish: strong baselines, held-out evaluation and explicit controls.

My direction is the evaluation and fidelity of language-model systems: whether a model, an agent or a policy behaves the way its results say it does. The projects below are where I practise that method, on reinforcement learning, recurrent networks and human-behaviour data.

Research

Project What I did
SCRES
reinforcement learning · simulation
With Alexander Garrido: supply-chain simulation and Gymnasium/PPO experiments judged against matched comparators.
Start with the same-contract comparison.
Motor-RNN connectivity
computational neuroscience
Neuromatch team project, then my independent equal-plasticity control, which separates structural density from trainable connections.
Start with the control note.
Spectral analysis of cognitive labor
human coordination
My reanalysis of Andrade-Lotero and Goldstone's human-search experiment, with a past-only temporal audit. Predictive advantage is not established.
Start with the temporal audit.

Software

  • HeliOS: a collaborative educational RISC-V kernel in C, to which I contributed; architecture notes and QEMU smoke tests.
  • ContratIA Abierta: open-procurement data pipelines, APIs and traceable signals for human review, not automated allegations.
  • ChaosLab: double-pendulum dynamics with numerical checks and an interactive explanation.

How to read my repositories

Read the question and the stated limitations, inspect a saved result and its method, then follow the documented test path. The existing evidence can be inspected without refitting any model.

Private work is described in the CV and is not published: a retrieval prototype for LLM support routing, numerical-analysis coursework, and a model audit of marmot social networks with Adriana Maldonado-Chaparro.

Tools: Python · PyTorch · NumPy/SciPy · NetworkX · SimPy · Gymnasium · R · C · SQL · FastAPI · Git · pytest

Pinned Loading

  1. scres-ia scres-ia Public

    Supply-chain simulation and RL policy evaluation: matched comparators, held-out tests and documented limits

    Python

  2. nma-motor-rnn-connectivity nma-motor-rnn-connectivity Public

    NumPy motor-RNN experiments: paired seeds, held-out evaluation and an equal-plasticity control (Neuromatch)

    Python 1

  3. spectral-cognitive-labor spectral-cognitive-labor Public

    Human collective-search reanalysis: past-only features and invariant spectral vs coordinate representations

    Python

  4. HeliOS HeliOS Public

    Collaborative educational RISC-V kernel in C with QEMU runtime checks

    C 1

  5. secop-risk-alerts-co secop-risk-alerts-co Public

    Colombian open-procurement data pipelines and interpretable signals for human review

    Python

  6. chaoslab-double-pendulum chaoslab-double-pendulum Public

    Physics course project: double-pendulum simulation, energy checks and interactive scientific visualization

    JavaScript