Senior Backend & Systems Engineer with 4.5 years of experience specializing in Spring Boot microservices, distributed systems, and event-driven architectures (Apache Kafka, AWS, Kubernetes). Currently exploring and deploying Local LLMs, Agentic workflows, and MCPs (Model Context Protocols).
- π Location: Hyderabad, India
- πΌ Role: Senior Associate Consultant at Infosys (ex-Cognizant)
- π Currently building: Native SCM text editor tooling and local LLM runtime experiments
| Domain | Technologies & Frameworks |
|---|---|
| Languages | Java, Go, Lua, SQL, Bash |
| Backend & Architecture | Spring Boot, Microservices, REST APIs, Distributed Systems |
| Messaging & Streaming | Apache Kafka, Dead Letter Queues (DLQ) |
| Cloud & DevOps | AWS (EC2, ECS/EKS), Kubernetes (K8s), OpenShift, Docker, Terraform |
| AI & LLMs | Ollama, llama.cpp, Local LLMs (Llama 3.2, Qwen 2.5), Agentic Workflows, MCPs |
| Security & Auth | OAuth2, JWT, Step-up Authentication |
| Observability & Tools | Splunk, Datadog, Prometheus, Spring Boot Actuator, Git, Maven, Gradle |
π Mana-Royale
- Stack: Go, Ollama, Local LLMs
- Implements a local AI card game with bluffing mechanics powered entirely by local runtime models (no external cloud API dependencies).
- Uses custom modelfiles and prompt/context engineering to simulate dynamic opponent deception and psychological tactics.
ποΈ Karm
- Stack: Java, Spring Boot, CLI
- Dual-mode task and project tracking utility operating in both CLI and Spring Boot modes.
- Manages work breakdown items, execution progress, and storage of associated multimedia assets (images, videos).
- Personal portfolio and technical write-up showcase.
- Enterprise Gateway Modernization (Infosys): Built and maintained Lua/Spring Boot API gateways and microservices deployed to Kubernetes via Terraform on AWS; migrated legacy OAuth1 services to modern OAuth2/JWT security patterns.
- Agentic Automation: Designed Go-based agentic testing pipelines running local LLMs (Llama 3.2, Qwen) via Ollama on Linux environments.
- High-Throughput Ingestion (Cognizant): Engineered Kafka-based asynchronous pipelines on AWS EC2 processing 150β250MB dynamic spreadsheets, reducing legacy batch latency from 21 days down to 5 minutes with dead-letter queue resilience and Prometheus metrics.
- Email: soumen8492@gmail.com
- GitHub: @Soumen-No-Hako
- LeetCode: Profile
- LinkedIn: Profile
