Jhansi Bhaskarla

Build and deploy production AI agents across multiple platforms, including Amazon Bedrock AgentCore and Gemini Enterprise. Serve as technical lead for WEX's Gemini Enterprise adoption specifically, owning licensing, evaluation, and rollout, and partner with business teams to identify use cases and build agents using each platform's tools. Built LLM analysis pipelines for operational data — including a Jira Service Management ticket processor and a service request aggregator pulling from Oracle RightNow CX, GitHub, Google Chat, Google Drive, and Jira Cloud — replacing manual review processes with automated summarization. Designed RAG and document ingestion pipelines for internal AI assistants with validation, filtering, and safety guardrails, backed by pytest test suites and Dockerized deployments. Built a real-time approval routing integration surfacing 270+ daily financial records directly in chat, and a Google Workspace AI assistant using Chat and Calendar APIs for natural language scheduling and focus time management.

Used HuggingFace transformer models to extract structured data from biomedical literature, with outputs feeding into clinical research workflows across three teams. Designed and deployed ML pipelines for tumor classification across multi-omics data using ensemble methods, iterating on feature selection based on model performance. Reduced Neo4j query times by 30% through Cypher optimization and schema changes. Deployed pipelines on AWS with Snakemake and Nextflow.

Built Python and R pipelines for single-cell RNA-seq analysis using Seurat and Scanpy on GEO datasets, covering normalization, dimensionality reduction, clustering, and differential expression to characterize tumor microenvironments. Integrated tumor genomic profiles with clinical metadata to study intratumor heterogeneity as part of a cancer subtype classification study.