AI/ML Engineer · Data Analyst · Software Developer
Building intelligent systems from Guwahati, Assam — transforming data into decisions.
I'm a Computer Science student at Royal Global University, Guwahati — graduating May 2026. My focus lies at the intersection of machine learning, data engineering, and software development.
With hands-on experience at IOCL and NIELIT, I've built ML pipelines, computer vision systems, and FPGA circuits. Currently training LLMs at Medium and developing IndicSignAI — an accessibility tool for the Deaf and Hard of Hearing community.
I believe in writing clean, purposeful code that creates real-world impact.
A hybrid AI framework transforming Assamese text into Indian Sign Language (ISL) animations via multilingual translation, speech synthesis, and avatar rendering — built for the Deaf and Hard of Hearing (DHH) community.
A full-featured Library Management System with dual user roles — Users (donate, apply, return books) and Admin (full inventory & student management). Clean Tkinter GUI backed by MySQL.
Developed during the IOCL internship — a real-time monitoring system that continuously tracks multiple sensor metrics (temperature, vibration, pressure, humidity). Applies machine learning to forecast hazard levels and equipment maintenance needs with 84.6% accuracy, cutting analytical workflow time by 40%.
How multi-agent architectures are reshaping the internet — from single chatbots to coordinated teams of autonomous AI systems completing real-world tasks.
READ ARTICLE →A deep dive into why memory and context limitations are the real bottleneck in modern AI agents — and how MCP and persistent memory are addressing it.
READ ARTICLE →Implementing human-in-the-loop approval workflows using LangGraph — keeping AI systems safe, auditable, and under meaningful human control.
READ ARTICLE →Exploring why large language models hallucinate, the real-world consequences, and practical techniques to detect and reduce confabulation in production systems.
READ ARTICLE →Comparing OpenAI's GPT-5 and Anthropic's Claude Sonnet 4 — training approaches, reasoning depth, and real-world performance benchmarks.
READ ARTICLE →A practical walkthrough of multiple classification algorithms on identical data — surfacing which models are useful and why.
READ ARTICLE →
Open to ML Engineer, Data Scientist, and SDE roles at innovative tech companies.
Let's create something impactful.