Vishnu Vikas Nallamalli
Software Engineer with 2+ years of experience building scalable microservices and distributed systems in cloud-native environments. Expertise in Java (Spring Boot) and Python, designing and optimizing REST and gRPC APIs to support high-throughput applications with sub-300ms p99 latency through collaboration with product, design, and infrastructure teams. Proven track record of improving system reliability and performance through schema redesign, telemetry-driven monitoring, and automation to maintain 99.9% SLA.
Experience
Software Engineer
State Street
USA
Designed and operated Java (Spring Boot) and Python microservices supporting ~15K daily transactions and peak ~300 RPS across distributed financial systems, incorporating unit testing frameworks to ensure code reliability. Reduced P95 API latency by 28% and improved P99 latency from 420ms to 295ms through optimized indexing, connection pooling, and bounded thread-pool concurrency models. Re-architected PostgreSQL and SQL Server schemas across ~150GB transactional datasets (~40M+ rows), improving complex query performance by 35% and reducing downstream reporting latency in time-sensitive workflows. Implemented secure REST and gRPC APIs across 6+ microservices using OAuth2 and RBAC controls, reducing integration regressions by 30% while maintaining 99.9% SLA compliance and backward compatibility. Introduced centralized telemetry and SLO monitoring across distributed services, reducing median incident resolution time by 30% and improving proactive detection of cross-service degradation. Automated CI/CD pipelines using Jenkins and GitHub Actions with blue-green deployment and rollback safeguards, increasing deployment frequency from bi-weekly to 3 releases per week while reducing post-release regressions by 40%. Optimized asynchronous processing and bounded executor configurations, reducing timeout errors under peak load by 22% and stabilizing tail latency during high concurrency traffic spikes.
Software Engineer
Virtusa
India
Developed Java and Python microservices supporting ~8K daily transactions and peak ~250 RPS, ensuring SLA adherence through resilient service interaction, idempotent request handling, structured exception management, and comprehensive unit testing. Optimized relational schemas and stored procedures across ~90GB transactional datasets (~20M+ rows), improving query execution performance by 40% through indexing, execution plan tuning, and refined data modeling. Refactored tightly coupled services into modular, independently deployable components, reducing deployment time by 30% and improving release stability across distributed systems. Implemented asynchronous queue-based workflows processing ~6K events/day, increasing throughput by 30% and strengthening retry and backpressure handling to prevent service overload during traffic spikes. Automated build validation and regression testing workflows using Python and CI pipelines, reducing build failure rates by 30% and improving release predictability within Agile sprint cycles.
Education
University of Maryland, Baltimore County
Master of Science
Baltimore County Coursework: Generative AI & ML Systems, MLOps and Production ML Pipelines, Distributed Cloud Systems, Cloud-Native Architectures, Large-Scale Data Engineering, Experimentation & Causal Inference
Skills
PROJECTS
Distributed Real-Time Feature Store & ML Inference Platform
Built a Kafka-based feature ingestion pipeline processing ~20K events/min with sub-100ms feature retrieval latency and exposed RESTful APIs for serving online predictions Implemented version-controlled feature definitions and wrote unit tests to ensure training-serving consistency and support scalable real-time prediction workflows under concurrent load
Enterprise AI Copilot with RAG & Guardrails
Engineered a production-grade RAG system in Python achieving 90%+ retrieval precision on offline evaluation datasets with ~900ms average response latency Integrated embedding pipelines, vector search, structured LLM validation, and role-based access controls, and enforced CI/CD practices including automated testing and code reviews to ensure enterprise-safe outputs
Cost-Optimized LLM Inference Orchestration Platform
Developed a Python-based inference orchestrator implementing dynamic batching and prefix caching, reducing per-request token cost by 25% while maintaining sub-second response latency under variable traffic conditions.