Vikash Kumar

Backend and GenAI Engineer with ~3 years of experience designing scalable microservices, Agentic AI workflows, and hybrid search platforms. Proven record of shipping LLM inference pipelines from prototype to production — delivering ~40% latency improvement, ~30% cost reduction, and enterprise-grade reliability. Multiple company award recipient for ownership, cross-functional collaboration, and on-time delivery under tight deadlines.

Associate Software Engineer — Backend & GenAI Engineer - Mphasis Pvt Ltd - Bangalore, India

(2023-08)

• Architected and deployed scalable AI-powered backend microservices via Docker, Kubernetes, and AWS, powering enterprise content management and intelligent search for the NeoZeta platform.
• Built a semantic caching layer (FAISS + Redis) cutting redundant LLM API calls by ~40%, improving response time by ~30%, and reducing infrastructure cost at scale.
• Engineered a FAISS hybrid search pipeline (vector + lexical) delivering sub-second semantic retrieval; architected Global Search over files and metadata using Vertex AI Embeddings, RDF4J & SPARQL.
• Optimized LLM inference via caching, batching, and model-routing — achieving ~30% cost reduction and ~40% latency improvement in production.
• Migrated Keycloak to SDK-based architecture; integrated OIDC with Azure Active Directory for enterprise SSO; automated CI/CD pipelines, boosting delivery velocity by 15–20%.

B.Tech - Computer Science Engineering - Amity University, Kolkata (2018 - 2022)