A selection of our recent work across AI research, natural language processing, computer vision, and production ML systems.
Each project represents our commitment to rigorous research and production-quality engineering.
End-to-end conversational AI system with multi-turn dialogue management, entity extraction, and sentiment analysis. Deployed across 3 enterprise clients handling 50k+ daily interactions.
Real-time visual defect detection system for manufacturing. Uses custom-trained object detection models to identify product defects with 99.2% accuracy at production line speed.
Open-source framework for building reproducible ML pipelines. Handles data versioning, experiment tracking, model registry, and automated deployment with rollback capabilities.
Novel meta-learning approach for medical image classification with limited labeled data. Achieved state-of-the-art results on dermatology and radiology benchmarks with 10x fewer training samples.
Automated document understanding system that extracts structured data from unstructured documents. Handles contracts, invoices, and legal filings with high accuracy and minimal human oversight.
Neural radiance field (NeRF) pipeline for photorealistic 3D reconstruction from 2D images. Used for architectural visualization, digital twins, and virtual environment generation.