I focus on building scalable AI systems, specifically in Deep Learning, Information Retrieval, and Reinforcement Learning. I love bridging the gap between research papers and production code, and I'm highly interested in applying AI to solve complex scientific and real-world challenges.

Here are the projects I am most proud of:

Discovering counterexamples for mathematical conjectures on graphs using Proximal Policy Optimization (PPO).

• The Challenge: Applying Deep RL to explore complex graph structures and refute theoretical conjectures.
• Tech Stack: Python, PyTorch, OpenAI Gym, PPO Algorithm.
• Key Insight: Designed a custom environment and reward function to train agents that actively seek out conjecture violations, proving the viability of RL in theoretical mathematics.

A comparative analysis of retrieval pipelines across diverse datasets.

• The Challenge: Evaluating how different search architectures perform on the rigorous BEIR benchmark.
• Tech Stack: Python, PyTorch, HuggingFace Transformers, FAISS.
• Key Insight: Implemented and compared lexical (BM25), sparse (SPLADE), and dense retrieval systems to optimize the balance between search accuracy and computational efficiency.

Languages

Machine Learning & AI

Data & Engineering

Open to collaborations on AI research and applications.