sachin kumar

At Cubastion Consulting, I lead a 10+ person team building SSC’s national-scale recruitment and examination platform, supporting 3.8 million+ online applications in a single SSC CGL 2025 cycle and tens of millions of annual transactions.

I design Node.js and Spring Boot microservices for high-availability government workloads, and I’m building an in-house RAG chatbot using LangChain, hybrid retrieval, PostgreSQL/pgvector, and a local Llama 7B-class model through Ollama. I also contribute to fine-tuning, retrieval evaluation, grounding, abstention, and secure handling of confidential knowledge sources.

I co-developed a real-time deep-learning face-quality validation system for SSC candidate photo capture, combining CNNs, MTCNN, image-quality checks, liveness processing, and TensorFlow Lite edge inference. This work was published as a Springer chapter.

Previously, I delivered AWS serverless applications at ency Consulting, reducing infrastructure costs by approximately 20–30%, and built full-stack advertising, compliance-management, and kiosk applications using React, Angular, Node.js, Express.js, and MongoDB.