Tejaswini Guddeti

I'm an AI/ML Engineer with hands-on experience building and deploying production multi-agent LLM systems, RAG pipelines, and computer vision models.

My most recent project is a 4-agent clinical reasoning system built with LangGraph where ResearchAgent pulls live PubMed literature, SafetyAgent checks FDA drug safety data, DiagnosisAgent builds differential diagnoses, and SynthesisAgent resolves disagreements into one structured report. Validated on 13 rheumatology cases with 13 out of 13 accuracy against ACR and EULAR classification criteria. Running live at 0.08 USD per case with a 4-layer security pipeline including prompt injection detection and LLM-based input validation.

I also built a hybrid RAG pipeline combining BM25 and dense embeddings with cross-encoder reranking, evaluated on golden and adversarial query datasets using precision@k and recall@k.

Industry experience includes DINOv2 contrastive learning for industrial defect featurization at Akridata where KNN accuracy improved from 85% to 95%+, and radar-based metal classification at CircleX Enterprises validated on live hardware with 90%+ accuracy.

I'm looking for remote AI engineering roles where I can build production agentic systems, RAG pipelines, or LLM-powered applications that solve real problems. I'm most interested in health-tech, applied AI, and early-stage companies moving fast.