LLM budget control and cost governance for AI agents. Python library for token budgets, usage limits and guardrails for OpenAI, Anthropic, LangChain, LangGraph and agentic systems.

AI-powered video auto-clipper that extracts the most compelling moments from long-form videos and turns them into vertical shorts optimized for TikTok, YouTube Shorts, and Instagram Reels. Built with Flask, FFmpeg, and Whisper.

TypeScript LLM Abstraction Layer for Multi-Provider AI Systems. Multi-provider routing, fallback chains, USD cost gating, capability factories, tool-use security.

A deterministic, safety-first microservice for parsing free-form bKash customer complaints (English, Bengali, Banglish) into structured triage data. Built with FastAPI, featuring a hybrid rule/LLM engine, local NLP normalization, and deep input/output guardrails.

A production-ready CLI Chatbot featuring an advanced RAG pipeline with Hybrid Search, Cross-Encoder Reranking, and a Multi-Model Fallback Router (Anthropic/Ollama). It includes Smart Memory context compression and a native LLM-as-a-Judge evaluation framework.

Self-hosted multi-agent assistant on Google ADK + Gemini 2.5 — FastAPI backend (agent-as-tool coordinator with ~22 tool specialists), Next.js voice UI, hardened auth, and an LLM fallback + completion-verifier resilience layer.

Hybrid SAP ticket routing: Rule Engine → TF-IDF → LLM fallback. Three-layer decision system with zero unnecessary API calls.

A production-ready, fault-tolerant multi-agent orchestrator built with LangGraph, FastAPI, and React. Features true async parallel task execution and dynamic open-source model fallback chains.