I am an engineer focused on delivering scalable backend services and Generative AI systems. My experience centers on integrating LLMs into reliable applications, with a hands-on approach to clean architecture, low latency, and advanced workflow orchestration (RAG, multi-agent frameworks, and multi-tenant systems).
Key capabilities I have built and delivered solutions around:
- Multi-Agent Orchestration: Implemented adaptive AI backends using frameworks like LangGraph to manage sequential reasoning, tool-calling for asynchronous CRUD operations, and multi-tenant persona state management.
- Conversational & Voice AI Systems: Built end-to-end voice and text-based systems, incorporating ASR (Sherpa-ONNX), multi-turn LLM control, RAG-enhanced reasoning, and real-time TTS for dynamic, low-latency interactions.
- Intelligent Data Pipelines & RAG: Developed unified systems for multimodal data processing, including semantic buffer aggregation, speaker diarization, and hybrid document retrieval using vector databases.
- High-Performance Backend: Established system scaling, reliable asynchronous task queues, and data consistency using modern Python frameworks (FastAPI), SQLAlchemy, PostgreSQL, Redis, and Celery within containerized environments.
- Backend & Core: Python (FastAPI), PostgreSQL, SQLAlchemy, Redis, Celery, MinIO, Docker, Docker Compose, Pytest.
- Generative AI: LangGraph, LangChain, RAG Architectures, Vector Databases (Qdrant, PGVector), LLM Orchestration (Groq, Gemini), ASR & Diarization (Sherpa-ONNX).
- Cloud & Tools: Google Cloud Platform (GCP), Azure, Twilio, ElevenLabs, Heygen, KlingAI, Firebase.
| Platform | Link |
|---|---|
| linkedin.com/in/kushalraga20 | |
| GitHub | github.com/ku5ha1 |
| kushalraga20@gmail.com |