Navod Peiris

@NavodPeiris · User

GitHub profile ↗ · Compare

Software Engineer | Maintainer

Colombo, Sri Lanka97 followers62 repositories

Repositories

NavodPeiris/agentic-avatars

Zero-Infrastructure Lip-Synced 3D avatar components for AI voice agents. This library is built with React Three Fiber and AI providers

★ 20TypeScriptForks 0

NavodPeiris/gsplat-talkinghead

LAM Gaussian-Splat Avatars for AI voice agents. Realistic avatars for interview/support/sales agents.

★ 1TypeScriptForks 0

NavodPeiris/airllm-cpu-inference-support

AirLLM makes it easy to infer large models through model sharding. This fork adds CPU support and single partition model loading and repartitioning.

★ 4Jupyter NotebookForks 1

NavodPeiris/speechlib

Speechlib is a library that unifies speaker diarization, transcription and speaker recognition in a single pipeline to create transcripts for audio conversations with actual speaker names and time tags. This library also contains audio preprocessor functions.

★ 269PythonForks 26

NavodPeiris/grizzlars

High-performance DataFrame library written in C++ with Python bindings.

★ 13PythonForks 1

NavodPeiris/graphsense

graphsense is a library to train and infer graph embedding based code suggestion models

★ 9PythonForks 1

NavodPeiris/DataFrame

C++ DataFrame for statistical, financial, and ML analysis in modern C++ with pybind11 support

★ 0Forks 0

NavodPeiris/MobileNet_96x96_greyscale_weights

weights of MobileNetV1 and MobileNetV2 trained on greyscale images. supports 96x96 image inputs only. Useful for developing models for Edge devices like Android, IOS and Microcontrollers.

★ 7Forks 0

NavodPeiris/BookVerse

BookVerse is a website which allows users to read, like, review, download, publish and buy books. We also offer free books and users can publish their own books and sell them. We want to give book lovers access to books worldwide and encourage authors to publish and make money.

★ 0JavaScriptForks 0

NavodPeiris/DoClasiq

A document classification and data extraction Web App for company documents. Classification is done using a fine-tuned version of LayoutLMv2ForSequenceClassification from HuggingFace Transformers library. Rule-based structured data extraction is done using predefined configuration for each class of documents.

★ 1Jupyter NotebookForks 0

NavodPeiris/bizsuit

E-Commerce data analytics and dashboarding app with self-serviceable model training and inference for churn prediction, working capital optimization and recommendation generation. Additionally, dashboarding is provided through Grafana for sales performance and customer monitoring

★ 0PythonForks 0