Harsh Kumar profile photo 1 on Tal

Harsh Kumar

Sr. Software Engineer @ Samsung R&D | Backend & Distributed Systems | Java | Spring Boot | AWS | Kubernetes

Bengaluru

4+ years of work experience

Samsung R&D Institute India - Bangalore
Mar 20233 yrs 8 mosPresent

Project: Bixby ASR — Automatic Speech Recognition – Redesigned the ASR inference serving architecture by integrating NVIDIA Triton with the Java-based ASR Engine and evaluating in-process and vLLM sidecar deployments, achieving up to 3x faster batch inference, 5% lower streaming latency, and 2x concurrent request capacity. – Integrated an internally trained SLLM into the production serving architecture, improving time-to-first-token (TTFT) by 8% while maintaining the existing Java-based inference workflow. – Architected a high-scale metadata platform for 80M+ audio samples, designing optimized PostgreSQL schemas and GIN indexes with Elasticsearch integration to enable efficient dataset discovery, filtering, and retrieval. – Led ASR model deployment and lifecycle management on AWS EKS, driving improvements across model release pipelines, deployment automation, and operational reliability for production ML workloads. – Automated model release workflows using Python, Shell, CI/CD, and GitOps, reducing manual deployment and release effort by 50% through reproducible, version-controlled deployments.
Samsung R&D Institute India - Bangalore
Mar 20231 yr 11 mosJan 2025

Project: Confidential B2C & B2B Communication Platform – Led the end-to-end migration of 12 Spring Boot microservices from AWS EC2 to EKS, containerizing services with Docker and Helm and implementing zero-downtime deployments, reducing deployment time to 15 minutes. – Designed and developed scalable Spring Boot REST APIs supporting real-time 1:1 and group communication workflows, integrating backend services with React/Angular frontends and collaborating across application and infrastructure layers. – Reduced backend API latency by 30% for 100K+ concurrent users by optimizing latency-critical workflows with Kafka for asynchronous event processing and Redis for distributed caching. – Built production CI/CD infrastructure using GitHub Actions and ArgoCD, enabling automated Kubernetes rollouts and reducing rollback time by 70% through declarative GitOps workflows. – Designed and deployed a production observability platform using Prometheus, Grafana, Loki, Tempo, and Mimir, processing 1M+ weekly logs and metrics and reducing issue identification time by 50%. – Improved automated test reliability and environment isolation by implementing post-test cleanup mechanisms and object-oriented test design patterns, reducing environment setup time by 20% and test flakiness by 15%.
HashedIn by Deloitte
Sep 20226 mosFeb 2023

Bengaluru, Karnataka, India

PROJECT: AR-based Food Menu Application - Architected the frontend of an AR-based food menu application using React Native, building reusable UI components and integrating them with Unity to deliver an interactive real-time 3D user experience. Integrated frontend workflows with AWS CloudFront and a MySQL backend to optimize 3D asset delivery, improve content availability, and support faster rendering of AR experiences. - Collaborated across frontend, backend, and Unity development workflows to improve integration quality and reduce friction between UI and 3D asset pipelines.
HashedIn by Deloitte
Jan 20228 mosAug 2022

Bengaluru, Karnataka, India

- Contributed to software development initiatives across application development, backend integration, and product delivery while building strong fundamentals in scalable engineering practices. - Developed production-oriented engineering skills across API integration, database-backed applications, debugging, version control, and cross-functional collaboration.

Harsh’s AI Stack

claude code
claude code
codex
codex
gemini
gemini
Cline
Cline

Education

Bhagwan Parshuram Institute of Technology

BTech - Bachelor of Technology, Computer Science

Bhagwan Parshuram Institute of Technology

20184 yrs2022

Join Harsh and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Harsh Kumar on Tal

Headline

Sr. Software Engineer @ Samsung R&D | Backend & Distributed Systems | Java | Spring Boot | AWS | Kubernetes

About

I build backend platforms that scale, ship reliably, and make engineering teams faster. I’m a Senior Software Engineer at Samsung R&D Institute, Bengaluru, with hands-on experience across backend engineering, distributed systems, DevOps, cloud infrastructure, CI/CD automation, observability, and data platforms. My work sits at the intersection of product engineering and platform reliability. I’ve built and optimized systems that process 80M+ audio samples, migrated 12 critical Spring Boot microservices from EC2 to AWS EKS, enabled zero-downtime deployments, reduced rollback time by 70%, improved API latency by 30%, and helped platforms reliably support 100K+ concurrent users. Technically, I specialize in: Backend Engineering: Java, Spring Boot, REST APIs, gRPC, microservices, Kafka, Redis Cloud & DevOps: AWS, EKS, Kubernetes, Docker, Helm, ArgoCD, GitHub Actions, GitOps Data & Storage: PostgreSQL, MySQL, Elasticsearch, Redis, schema optimization, GIN indexing Observability: Prometheus, Grafana, Loki, Tempo, Mimir, alerting, distributed tracing Engineering Ownership: architecture, deployment automation, performance optimization, production reliability What differentiates me is that I don’t just write services, I take ownership of the full engineering lifecycle: architecture, implementation, testing, deployment, monitoring, and reliability. I enjoy solving ambiguous backend problems, simplifying complex systems, and building tooling that improves developer velocity. I’m currently targeting SDE-2/Senior Software Engineer/Backend Engineer/Platform Engineer roles where I can build scalable backend systems, own high-impact end-to-end services, and contribute to engineering cultures that value ownership, craftsmanship, and reliability.

Location

Bengaluru

AI stack

claude code

codex

gemini

Cline

Work experience

Senior Software Engineer at Samsung R&D Institute India - Bangalore, Mar 2023 - Present · 3 yrs 8 mos, Project: Bixby ASR — Automatic Speech Recognition – Redesigned the ASR inference serving architecture by integrating NVIDIA Triton with the Java-based ASR Engine and evaluating in-process and vLLM sidecar deployments, achieving up to 3x faster batch inference, 5% lower streaming latency, and 2x concurrent request capacity. – Integrated an internally trained SLLM into the production serving architecture, improving time-to-first-token (TTFT) by 8% while maintaining the existing Java-based inference workflow. – Architected a high-scale metadata platform for 80M+ audio samples, designing optimized PostgreSQL schemas and GIN indexes with Elasticsearch integration to enable efficient dataset discovery, filtering, and retrieval. – Led ASR model deployment and lifecycle management on AWS EKS, driving improvements across model release pipelines, deployment automation, and operational reliability for production ML workloads. – Automated model release workflows using Python, Shell, CI/CD, and GitOps, reducing manual deployment and release effort by 50% through reproducible, version-controlled deployments.

Software Engineer at Samsung R&D Institute India - Bangalore, Mar 2023 - Jan 2025 · 1 yr 11 mos, Project: Confidential B2C & B2B Communication Platform – Led the end-to-end migration of 12 Spring Boot microservices from AWS EC2 to EKS, containerizing services with Docker and Helm and implementing zero-downtime deployments, reducing deployment time to 15 minutes. – Designed and developed scalable Spring Boot REST APIs supporting real-time 1:1 and group communication workflows, integrating backend services with React/Angular frontends and collaborating across application and infrastructure layers. – Reduced backend API latency by 30% for 100K+ concurrent users by optimizing latency-critical workflows with Kafka for asynchronous event processing and Redis for distributed caching. – Built production CI/CD infrastructure using GitHub Actions and ArgoCD, enabling automated Kubernetes rollouts and reducing rollback time by 70% through declarative GitOps workflows. – Designed and deployed a production observability platform using Prometheus, Grafana, Loki, Tempo, and Mimir, processing 1M+ weekly logs and metrics and reducing issue identification time by 50%. – Improved automated test reliability and environment isolation by implementing post-test cleanup mechanisms and object-oriented test design patterns, reducing environment setup time by 20% and test flakiness by 15%.

Software Developer at HashedIn by Deloitte, Sep 2022 - Feb 2023 · 6 mos, Bengaluru, Karnataka, India, PROJECT: AR-based Food Menu Application - Architected the frontend of an AR-based food menu application using React Native, building reusable UI components and integrating them with Unity to deliver an interactive real-time 3D user experience. Integrated frontend workflows with AWS CloudFront and a MySQL backend to optimize 3D asset delivery, improve content availability, and support faster rendering of AR experiences. - Collaborated across frontend, backend, and Unity development workflows to improve integration quality and reduce friction between UI and 3D asset pipelines.

Software Developer at HashedIn by Deloitte, Jan 2022 - Aug 2022 · 8 mos, Bengaluru, Karnataka, India, - Contributed to software development initiatives across application development, backend integration, and product delivery while building strong fundamentals in scalable engineering practices. - Developed production-oriented engineering skills across API integration, database-backed applications, debugging, version control, and cross-functional collaboration.

Education

BTech - Bachelor of Technology, Computer Science, Bhagwan Parshuram Institute of Technology