A local tool for syncing, searching, and exporting AI coding sessions across development tools, with a terminal UI and semantic search.

A local-first SQL IDE for flat files that combines DuckDB, FastAPI, Vite, and LangGraph for semantic query generation.

A text humanness scorer and reinforcement-learning pipeline that turns model evaluations into a reward function for Gemma fine-tuning.

Experiments with keyed text watermarking, statistical detection, model behavior, calibration, and the effect of editing on the signal.

An entity-resolution evaluation comparing dense embeddings with BM25 and Matryoshka Representation Learning for efficient retrieval at scale.

A Llama 2 7B model fine-tuned on 60,000 bilingual English-Gujarati pairs for low-resource language use cases.

A multi-model retrieval-augmented system for interacting with autonomous-vehicle video streams, built for the LlamaIndex RAG-a-thon.

Statistical and machine-learning models for estimating wind-farm power curves and supporting energy-system optimization.

Classification models for predicting churn, appetency, and up-selling behavior for a mobile network operator.

Multivariate quality-control analysis for an industrial forging process using principal components and control charts.