At Scalric, I design, build, and deploy LLM-powered AI agents and RAG pipelines for production applications through RESTful APIs. I test and refine prompts and tool-use logic using user feedback to improve accuracy and reliability.
At Intrainz, I architected RAG systems with LangChain and FAISS or Pinecone, improving LLM response accuracy by 35%. I also fine-tuned BERT and T5 models for text classification, NER, and summarization, achieving up to 91% accuracy.
On my Multi-Agent AI Workflow project, I used AutoGen and LangChain to enable agents to collaborate and delegate tasks; persistent memory and planning logic improved task completion accuracy by 28% over single-agent baselines. I also worked as a Data Analyst Intern at EliteTech, using Python, SQL, and Excel to analyze business datasets and building Power BI dashboards.
