Hi, I'm Ebenezer!

Engineer working on AI, with experience in reinforcement learning. I build things and test what works.

GitHub Activity

Live Data

6,188 contributions in the last year

Less

ebenezergelo/roadster_flip

Teach an RL agent to flip a car in mid-air and land it.

ebenezergelo/mujoco_ant

Teaching Mujoco Ant to walk with PPO

Updates

July 2026Joined Melio AI as a Data Scientist
July 2026Completed MSc in Computer Science with distinction from Wits University
Nov 2025Came 2nd on McKinsey's Tech2Impact hackathon
Jun 2025Joined LiGHT Lab as part of the Summer@EPFL program
Apr 2025Talk on "Technical Alignment in Autonomous Agents" at AI-ZA Meetup
Apr 2025Started AI Safety Wits

Apr 2025Facilitator for "Transformative Artificial Intelligence" course by AISSA
Dec 2024Attended the Condor Camp South Africa
Jan 2024Started my MSc at Wits University, working in RAIL Lab

Publications

Redistribution-based Cost Inference Improves Sparse Safe Offline RL
Ebenezer Gelo*, Geraud Nangue Tasse, Steven James, Benjamin Rosman
1st IJCAI Workshop on Safe Physical AI (SPAI) @ IJCAI-ECAI 2026. Aug 2026
arXiv PDF Code
CORDA: A Benchmark for Hierarchical Harm-Centric Moral Reasoning in Large Language Models
Siddarth Singh*, Victoria Williams, Simon Rosen, Ebenezer Gelo, and
Fourth International Workshop on Value Engineering in AI (VALE) @ IJCAI-ECAI 2026. Aug 2026
arXiv PDF
Predicting Recurrence After Hepatocellular Carcinoma Resection Using Deep Learning on Gross Tumor Images
Stéphanie Gonvers*, Ebenezer Gelo*, André Bubna Hirayama, and
IHPBA, Abstract. Mar 2026
Code
MoralityGym: A Benchmark for Evaluating Hierarchical Moral Alignment in Sequential Decision-Making Agents
Simon Rosen*, Siddarth Singh, Ebenezer Gelo, and
International Conference on Autonomous Agents and Multiagent Systems (AAMAS). Feb 2026
arXiv PDF Code

Projects

Inferring Prognosis for Macroscopic Hepatocellular Carcinoma

Using multi-modal deep learning to predict disease progression in Hepatocellular Carcinoma (HCC) patients by integrating clinical data and medical imaging.

# healthcare# deep-learning# multi-modal-ml

Redistribution-based Cost Inference for Safe Offline RL

Integrating expert feedback into offline RL to enhance decision-making in safety-critical environments while minimizing risks associated with live trials.

# offline-rl# rlhf# safe-ai# constrained-rl

Learning World Value Functions(WVFs) without Exploration

Exploring how offline RL algorithms learn goal-oriented representations - World Value Functions (WVF), emphasizing the importance of dataset size and diversity.

# offline-rl# deep-learning# goal-oriented-rl

Hi, I'm Ebenezer!

GitHub Activity

Updates

Publications

Redistribution-based Cost Inference Improves Sparse Safe Offline RL

CORDA: A Benchmark for Hierarchical Harm-Centric Moral Reasoning in Large Language Models

Predicting Recurrence After Hepatocellular Carcinoma Resection Using Deep Learning on Gross Tumor Images

MoralityGym: A Benchmark for Evaluating Hierarchical Moral Alignment in Sequential Decision-Making Agents

Projects

Inferring Prognosis for Macroscopic Hepatocellular Carcinoma

Redistribution-based Cost Inference for Safe Offline RL

Learning World Value Functions(WVFs) without Exploration