Ajit Sivakumar

Research Interests

• Reinforcement learning
• Deep learning
• Reward modeling
• Mechanistic interpretability

Experience

Research Engineering Intern: d_model

March 2026 – Present

Engineering reinforcement learning environments with a team of OpenAI, Anthropic, Deepmind, and MATS alum.

Software Engineer Intern: Superfocus.ai

September 2025 – March 2026

Developed conversational memory modules with temporal decay for embedded LLMs.

Research Intern: Rebellion Research

June – September 2025

Built a python-based machine learning pipeline using XGBoost and SHAP to analyze factor importance in ETFs.

Education

B.A. Mathematics and Computer Science, New York University

Coursework

• Data Structures
• Basic Algorithms
• Computer Systems Organization
• Linear Algebra
• Calculus III (Multivariable Calculus)
• Discrete Mathematics
• Analysis
• Theory of Probability
• Parallel Computing
• Object Oriented Programming
• Applied Internet Technology

Skills

Programming

• Python
• JavaScript
• TypeScript
• C++/C
• Java
• SQL
• x86 Assembly
• HTML & CSS

Machine Learning

• TensorFlow
• PyTorch
• scikit-learn

Web & Tools

• React & Next.js
• Node.js
• Flask
• Git
• LaTeX
• Docker

Mathematics

• Probability Theory
• Statistical Modeling
• Optimization

Projects

Where2Meet

Flask, React

A full-stack web application that helps users find optimal meeting locations based on their preferences and constraints.

SAIL - Financial Document Analyzer

Flask, React, LangChain, Selenium, TensorFlow

Web application that uses NLP techniques to extract and analyze key financial metrics from SEC filings and reports.

Movement Classification System

OpenCV, TensorFlow

Real-time movement analysis system using computer vision and machine learning for athletic performance evaluation.

Portfolio Website

Next.js, TypeScript, React

My personal website.

Contact

You can reach me at ajit [dot] sivakumar [at] gmail [dot] com, or connect with me on LinkedIn.