Partik Malasi

About Partik Malasi

I am Partik, a results-driven Software Engineer passionate about building scalable, high-performance, AI-powered systems. Currently pursuing my B.Tech in Computer Science and Engineering at Kota, I bring strong expertise in Golang, C++, Python, Java, and C, along with a modern full-stack skill set including React.js, Node.js, Express.js, Next.js, and Tailwind CSS.I am currently working as a Software Engineer Intern at Zomato, contributing to the Nugget AI team, where I’m building AI agents for customer support, scalable support automation, ticketing workflows, and high-impact customer experience products using Golang and GenAI.Previously, I worked as an SDE Intern at Amazon in the Manifest (Ship Tech) team, where I developed the GenAI Template Management System (ITMS), a large-scale platform enabling LLM-powered template creation, validation, and macro automation. I designed a multi-region knowledge base architecture supporting 700+ production templates using DynamoDB, S3, and Amazon Kendra, and integrated a Retrieval-Augmented Generation (RAG) pipeline to deliver knowledge-grounded GenAI responses at scale.My past experience also includes developing a WebSocket-based LLM prompt validation toolkit at Cisco, building full-stack cyber intelligence platforms at Cyble, and creating a production-grade e-commerce platform from scratch at DhagaKart.With strong foundations in system design, data structures and algorithms, backend architecture, and distributed systems, I thrive on solving complex engineering problems and delivering real-world impact through clean, scalable code.

Experience

1. Software Engineer Intern

   Zomato

   Jan 2026 — Present · Gurugram, IN

   Working on end-to-end real-time Voice AI pipelines (STT → LLM → TTS) supporting production-scale automated voice interactions. Integrated OpenAI as a provider in the LLM generator factory, enabling multi-model support within the Voice AI system. Improved system reliability by fixing ungraceful shutdowns in Firefly streams, preventing resource leaks during active calls. Implemented handling for WebSocket abnormal closure (1006) to eliminate dangling goroutines and strengthen concurrency safety. Resolved Redis-based session recovery issues during EC2 rotations, ensuring ongoing calls remained stable during deployments.

Education

• Kendriya Vidyalaya

  Senior Secondary Examination

• Indian Institute of Information Technology Kota

  Bachelor of Technology - BTech, Computer Science

• Kendriya Vidyalaya

  Secondary Examination