Devesh Singh

I’m a Software Development Engineer at FinMechanics, focused on building high-throughput, fault-tolerant backend platforms. I architected a real-time accounting engine processing 100k+ entries/day across 15+ banking clients using Kafka-based event driven postings with idempotent consumers for exactly-once semantics under retry and failover.

I also engineered a lock-free hash-partitioned parallel execution engine, reaching 8k+ entries/sec versus overnight batch processing. To keep systems adaptable, I replaced hardcoded business rules with a SpEL expression engine, enabling runtime configurability and eliminating deployment cycles.

Beyond core processing, I built integration and reporting capabilities: a REST Executor framework standardized across 10+ external systems, and an integration pipeline using Apache Camel polling, Hazelcast queues, and DLQ isolation for guaranteed delivery. On the reporting side, I embedded a Groovy Shell scripting runtime for custom report authoring without deployments, with JSON templates scheduled via Quartz and automated email delivery across 15+ banking clients.

My project work reflects the same commitment to correctness and reliability. I implemented Raft Consensus with leader election, snapshotting, and crash-safe WAL; built a Saga Orchestrator with WAL semantics and deterministic idempotency keys; and developed a Transactional Outbox Service ensuring atomic business+event writes for exactly-once delivery. I also designed a DAG-based Workflow Orchestration Engine with parallel execution and context propagation using Virtual Threads and benchmark-driven performance improvements.