At BuzzyBrains Technologies Pvt. Ltd., I built a production RAG pipeline that improved retrieval accuracy by 30% and cut latency by 20%. I also developed LangGraph workflows with an LLM gateway and safety guardrails.
I fine-tuned Whisper and Gemma models and developed audio-classification models, taking the pipeline from preprocessing through deployment. I also designed Django and FastAPI backends and Redis caching that reduced API response latency by 70% under high-load scenarios.
On projects including Yes Mentor and Asha, I built AI services for resume generation and document-based knowledge retrieval. My work includes embedding pipelines, vector search, prompt-injection defenses, and PII redaction.
