Karan Kumar
officialkaran39@gmail.com · +91 7903214466
github.com/09karankr · linkedin.com/in/karan-kumar-122787226
Professional Summary
Software Engineer at Capgemini with hands-on experience in GCP, Docker, Kubernetes, and full-stack JavaScript/TypeScript. Currently learning the modern AI/ML stack — PyTorch, Hugging Face, LangChain, and RAG-based LLM applications. Strong CS fundamentals with 400+ DSA problems solved on CodeChef and LeetCode.
Technical Skills
- Languages: Python, Java, C++, JavaScript, TypeScript, SQL
- AI / ML: PyTorch, Hugging Face, LangChain, RAG, FAISS, OpenAI API, Pandas, NumPy, Prompt Engineering
- Web: React.js, Next.js, Node.js, Express.js, REST APIs
- Cloud: AWS, GCP, GCS, BigQuery
- DevOps: Docker, Kubernetes, Jenkins, Terraform, Ansible, Prometheus, Grafana, CI/CD
- Databases: MySQL, BigQuery
- Tools: Git, GitHub, Linux, VS Code
Professional Experience
- Deploy and manage scalable containerized applications on GCP, leveraging GCS buckets for structured and unstructured data storage across services.
- Built backend-centric Next.js services using SSR and API routes, integrating SQL and BigQuery to process large-scale datasets and improve API response latency by ~30%.
- Containerized applications with Docker, enabling consistent deployments and smoother release workflows across environments.
- Completed 8+ weeks of structured Core Java training, applying OOP, collections, and exception handling in 2+ mini-projects.
- Designed CI/CD pipelines using Git, Jenkins, and Docker, automating build and deployment for 3+ services and integrating Java, Python, and DevOps tools.
Key Projects
Python, PyTorch, LangChain, Hugging Face, FAISS, OpenAI API
- Built a Retrieval-Augmented Generation (RAG) chatbot ingesting 100+ documents, using sentence-transformer embeddings and FAISS vector search for sub-second semantic retrieval and citation-backed answers.
- Fine-tuned a Hugging Face transformer on domain QA pairs using PyTorch with LoRA adapters, reducing training memory footprint by ~60% on a single GPU.
- Orchestrated multi-step LLM workflows using LangChain agents and tool-calling, enabling structured query decomposition, function execution, and grounded responses via the OpenAI API.
AWS, Jenkins, Docker, Kubernetes, Terraform, Ansible
- Built an AWS-based CI/CD pipeline automating build, test, security scans, and Kubernetes deployments for containerized microservices.
- Configured a 10+ stage Jenkins pipeline integrating GitHub, Maven, SonarQube, Nexus, Docker, and Kubernetes; provisioned infrastructure with Terraform and Ansible, reducing setup time by ~80%.
- Implemented Trivy container security scanning and Prometheus-Grafana monitoring, achieving a 90% reduction in deployment effort.
Node.js, Express, MongoDB, JWT, Zod
- Built 10+ REST APIs for signup/signin and course CRUD with modular routing and Zod-based input validation, supporting Admin and User flows end-to-end.
- Implemented JWT authentication and bcrypt password hashing, securing user credentials and role-based access across all protected endpoints.
React, TypeScript, Node.js, Express, WebRTC, Socket.io
- Built a real-time video conferencing app with low-latency peer-to-peer audio/video streaming over WebRTC and a room-based architecture supporting multi-user meetings.
- Integrated real-time chat and join/leave notifications via Socket.io to improve meeting collaboration.
Education
Certifications
- AWS Certified Cloud Practitioner (CCP)
- Google Cloud Innovator
- GitHub Foundations Certification
- Docker Hands-On: Containerization Fundamentals