Madhur Singhal

Backend engineer with 5 years of experience building distributed systems and cloud infrastructure at Salesforce. Experienced in Java, distributed caching, asynchronous processing, event-driven systems, gRPC services, service-to-service security, concurrency, and infrastructure automation. Built production systems under strict latency, memory, scalability, and reliability constraints, with ownership spanning design, implementation, deployment, and production support.

Promoted to Member of Technical Staff (MTS) in H2 FY24.

Member of Technical Staff at Salesforce (2022-01 – Present)

• Diagnosed a production incident where filesystem-backed content loading took ~36s across 2,600 items in 81 namespaces.
• Redesigned monolithic caching into namespace-level lazy loading, reducing cache invalidation scope and avoiding reload storms while operating under a 64 MB JVM heap constraint.
• Profiled JVM memory usage and identified cache growth as a scalability bottleneck; separated lightweight metadata from large content, keeping metadata in-memory and large content in a shared distributed cache.
• Used ConcurrentHashMap and computeIfAbsent to coordinate concurrent local cache population and prevent duplicate resource loading.
• Root-caused a 12× performance regression using Splunk, identifying sequential namespace loading as the primary bottleneck.
• Parallelized namespace loading using Java Parallel Streams and ConcurrentHashMap, reducing page-load latency from 4.84s to 1.49s (3.2×).
• Designed an asynchronous export pipeline using QPID, HBase, and object storage, decoupling large report generation from synchronous APIs and reducing request latency to <10ms for exports of up to 50K records.
• Optimized pagination and batching and introduced bounded worker concurrency, improving export throughput by 4.1× while protecting shared production resources.
• Compressed and streamed large report artifacts to object storage, reducing a ~54 MB test export to a ~1.97 MB downloadable artifact.
• Designed a concurrent KPI aggregation engine using managed thread pools, executing up to six independent aggregate queries in parallel with dynamically generated queries and filters.
• Designed an event-driven ingestion platform routing events to internal storage, Salesforce Platform Events, and external analytics providers through configurable, feature-gated pipelines.
• Implemented scalable processing patterns including optimistic locking and cursor-based pagination for high-volume event processing.
• Built production Java platform services communicating through gRPC and REST, including reusable health-check infrastructure, dependency-specific probes, retries, response aggregation, and secure object-storage operations.
• Built reusable storage infrastructure supporting upload, download, delete, metadata operations, multipart upload, and presigned URLs, with backend storage abstractions decoupling storage providers from business logic.
• Designed blue-green and canary deployments, progressive delivery, rollback procedures, and feature-flag-based releases; implemented mTLS, HMAC authentication, authorization, and rate limiting.

Backend Engineer at 314e Corporation (2021-01 – 2022-12)

Built backend services for a multi-tenant healthcare SaaS platform.

• Built asynchronous backend services using Python, FastAPI, PostgreSQL, RabbitMQ, and Kafka for a multi-tenant healthcare SaaS platform.
• Developed runtime-configurable schema generation supporting tenant-defined data models.
• Integrated Elasticsearch for scalable search and containerized services using Docker for zero-downtime deployments.

B.Tech in Computer Science & Engineering – National Institute of Technology, Raipur