12 projects
nepal-gov-agent
Reference implementation: agentic RAG on Nepal government policy and legal documents. Offline-capable, Nepali + English, hybrid retrieval with citations.
agentensemble
Build, orchestrate, trace, and test multi-agent LLM systems in Python: ReAct, RAG, router, planner, workflow graphs, swarms, debates, and pipelines.
ragfallback
Prevents silent RAG failures — chunk quality, retrieval fallback, adaptive querying, and answer evaluation in one library.
ragnav
Hybrid structure-aware retrieval — BM25 + embeddings + document graph expansion. Runs fully offline, no API key required.
pyrochain
PyTorch + LangChain for agentic feature engineering: multimodal (text + image) feature extraction with LoRA adapters, validation agents, and conversation memory.
agentcare
Framework for healthcare voice-AI agents: front-desk booking, follow-ups, wellness check-ins, transcript extraction, and analytics.
lingo-nlp-toolkit
Unified NLP toolkit: one Pipeline API for sentiment, NER, classification, embeddings, QA and summarization on Hugging Face models, plus spaCy/NLTK text preprocessing.
askpandas
Ask questions about pandas DataFrames and CSV files in plain English: an LLM writes pandas code and AskPandas runs it in a restricted sandbox (Ollama, OpenAI, Mistral, Hugging Face or any OpenAI-compatible API).
trustpilot-scraper
Scrape public Trustpilot company reviews into plain Python dicts, CSV or JSON, using the page's embedded __NEXT_DATA__ JSON.
scrapeflow-py
Opinionated Playwright scraping workflow engine: retries, rate limiting, anti-detection, robots.txt compliance, Pydantic spec-driven and LLM-powered extraction, sessions, and metrics.
socialmediaextractor
Extract a website's social media profile links (X/Twitter, LinkedIn, Instagram, Facebook, YouTube, TikTok, GitHub, Threads, Bluesky), skipping share buttons.
toxic-comment-classifier
Pre-trained BiLSTM classifier that scores comments for six kinds of toxicity (toxic, severe_toxic, obscene, threat, insult, identity_hate). Works offline, one line to use.