About

About

Electronics & Communication at IIT Patna, B.Tech + M.Tech. My focus is physical AI: learned robot policies, vision-language-action models, and world models that let a robot predict before it acts. I came to it from the hardware side — competition robots, a drone that lands on three rotors, a robot that drives to a spoken goal — so I care about what survives contact with the real world.

§ 01

How I work

01

The edge survives alone

Networks drop. Anything on the robot has to degrade into something safe without help.

02

A system must explain itself

Append-only ledgers and journaled jobs are what make autonomy deployable, not overhead.

03

Build the layer below once

Robotics time disappears into glue. A real substrate pays back across everything above it.

04

Privacy from physics

A camera you promise not to misuse is still a camera. Radio and electric fields keep the promise for you.

Match play at DD Robocon 2025.

Education

B.Tech + M.Tech (Integrated) — Electronics & Communication Engineering

Indian Institute of Technology Patna · 2023 — 2028

Signal processing, computer architecture, embedded systems, control theory, and machine learning.

§ 02

The shape of every system here

Real-time at the bottom, deliberation at the top.

Three tiers: Operator on the robot, Tactician arbitrating control, Strategist planning off the critical path. Observations flow up, commands flow down.L2Strategistseconds · may be slowMission planLanguage modelMemory ledgerL1Tactician100 ms · arbitratesTask managerPlannerAuthorityL0Operator1 ms · must not missSensorsSLAMControl loop
observations, up commands, down- - - timing contract