Research Projects

Our research work across inference optimization, computer vision, and quantitative finance.

V0 BATCH

Completed Research

Published

PyTorch native INT8 quantization API

INT8 tensor subclass for PyTorch enabling up to 4× memory reduction with optimized CUDA/Triton kernels. Merged into TorchAO.

Quantization PyTorch TorchAO
Read Report → GitHub →
Published

nano-vLLM

Educational LLM inference engine built from scratch. Covers PagedAttention, continuous batching, chunked prefill, and scheduling with detailed C++ implementations.

Inference vLLM LLM Serving
Part 1 → Part 2 → GitHub →
V1 BATCH

Current Research

In Progress

ViT Robustness via Occlusion

Improving Visual Transformer robustness using occlusion generators for better generalization.

Computer Vision ViT Robustness
In Progress

Sparse Frame Selector

Building sparse frame selection methods for faster video reasoning and efficient temporal understanding.

Computer Vision Video Efficiency
In Progress

Prediction Market Analysis

Using LLMs for prediction market analysis and probabilistic forecasting.

Quant LLM Markets
In Progress

Event-Based Trading Agents

Developing LLM-powered trading agents that react to market events and news.

Quant Agents Trading

Interested in collaborating on research? Contact us at daniel@aerlabs.tech or shubham@aerlabs.tech