abhishek kumar

Architected and deployed a production multi-agent LLM platform for drug discovery using LangGraph, with memory-augmented reasoning and evaluation/monitoring for regulated auditability. Built pharma R&D analytics and HTS workflows, including narrowing 3M compounds to 3,000 high-confidence hits via supervisor multi-agent reasoning.

Architected an enterprise RAG platform with full LLM and data-layer orchestration across business units, establishing standards for retrieval, embedding selection, and secure deployment on Kubernetes. Implemented hybrid retrieval across Pinecone and Elasticsearch and reduced infrastructure costs by ~90%, while adding evaluation, monitoring, and drift governance that reduced manual validation by ~4

Built scalable data quality pipelines to reduce manual validation overhead and improve downstream analytics reliability. Developed real-time and near real-time ingestion/ETL for medical and IoT sensor data using PySpark and AWS services, including production troubleshooting and mentoring for distributed processing best practices.

Contributed to a startup building data engineering pipelines for stock market analytics, including ingestion and backtesting workflows to support data-driven decisions. Refactored a monolith into a Python/AWS microservices architecture and built backend and frontend components for a poller application with CI/CD automation via Jenkins.

Developed and deployed machine learning models in Python and R, exposing them as scalable REST APIs using Django/RPlumber and MLOps containerization with Docker. Built data governance and data ingestion services for enterprise platforms using Django, Flask, PySpark, Swagger, and PostgreSQL to ensure compliant processing and traceability.

Earned a Bachelor of Engineering (B.E.) in Information Science & Engineering at Dayananda Sagar College of Engineering (VTU University) in Bengaluru, India, from 2011 to 2015.