Anshu Bhadani

A sophisticated clinician-patient matching engine powering a US healthcare SaaS platform serving multiple organizations simultaneously. Engineered for zero-downtime operation under HIPAA compliance constraints, this system combines the intelligence of semantic embeddings with the transparency of rule-based scoring — giving healthcare administrators both powerful automation and full auditability for regulatory compliance.

• →Architected dual-mode matching: embedding-based semantic similarity (Sentence Transformers + FAISS) paired with deterministic weighted scoring for transparent, explainable decisions
• →Designed ephemeral, per-tenant vector indexing strategy ensuring strict data isolation in multi-tenant environment — critical for healthcare data sovereignty requirements
• →Built comprehensive feedback capture loop logging match outcomes, creating foundation for supervised learning-to-rank with automated model retraining pipelines
• →Delivered production system processing real patient-clinician matching with sub-second latency while maintaining complete audit trails for compliance reviews

An intelligent document parsing platform that revolutionizes legal contract management by automatically extracting dates, events, and contractual obligations from unstructured legal documents. By combining precision rule-based parsing with sophisticated LLM-powered natural language understanding, this system reduced manual legal review cycles by 40% — translating to hundreds of billable hours saved for our legal enterprise clients.

• →Designed hybrid AI pipeline leveraging rule-based extraction for structured elements and LLM inference for nuanced legal language interpretation and obligation classification
• →Architected event-driven microservices architecture with RabbitMQ-based decoupling of ingestion, extraction, scheduling, and notification services for independent scaling and fault isolation
• →Led technical design for bidirectional CRM/ERP integration enabling seamless workflow automation and obligation tracking across enterprise legal operations software
• →Positioned solution for SOC 2 Type II and ISO 27001 certification readiness, implementing encryption at rest/transit, comprehensive audit logging, and role-based access controls for sensitive contract data

A white-label, enterprise-ready retrieval-augmented generation (RAG) platform empowering businesses to deploy document-grounded AI chat support systems with zero shared infrastructure. Each tenant operates in complete isolation with independent vector stores, dedicated embedding pipelines, and row-level database access controls — addressing the primary enterprise concern blocking AI adoption: data security and cross-tenant contamination risk.

• →Architected complete multi-tenancy with per-tenant vector store isolation, ensuring no possibility of data leakage between organizations using shared embedding models and retrieval infrastructure
• →Built comprehensive admin portal with real-time analytics dashboards, usage monitoring, document management, and fine-grained configuration controls for chunking strategies and retrieval parameters
• →Developed production REST API and embeddable JavaScript widget enabling seamless integration into existing customer support platforms, knowledge bases, and internal tools
• →Designed scalable architecture supporting enterprise adoption with horizontal scaling, caching strategies, and cost optimization through batch embedding processing and vector index compression