about · the longer version
i build systems that still make sense at 2am.
i'm avyukt — computer engineering at AMU (ZHCET), pre-final year. i work full-stack at a startup (stick), run my own company (continuum), ship apps, jump into hardware when the problem isn't only software, and i'm working toward an international conference paper (voltSense / ieee impact line—still in motion).
when a project needs it, i handle photography, editing, and graphics myself. lately i've been putting more of the build process on camera—short videos of the messy middle live on @avyukt_builds on instagram.
my founder thinks i'm self-obsessed. i'm choosing to read that as alignment on narrative. the rest of my time still goes to mobile + backend, agents, pr automation, and hardware spikes—esp32, wearables, iot—anything where i can't sleep until the signal path makes sense.
“the best feeling isn't when it demos—it's when it survives contact with reality.”
i've always learned by taking hardware apart and putting it back together—that turned into arduino builds and, later, the same instinct applied to backends and deploys: make the signal path obvious, then make it reliable.
hackathons became the default rhythm: ship something real by the deadline, learn what breaks, repeat. the big annual events in my city were already running long before i showed up—i wasn't the one who built that machine; i just learned to live inside it.
01
high school → the hackathon gravity well
i found hackathons the way some people find sports: loud rooms, impossible deadlines, strangers who became teammates overnight. the rule was simple—ship something by sunday. i started hitting every event i could reach—including the ones that pack a few hundred people into a hall that had been doing that for years before i arrived. robots, ar, tools that shouldn't have worked but did.
that's when it clicked: i didn't only want to build things. i wanted to build rooms where other people could build too.
national finalist
SIH 2025 · NeuroSense
top 105 / 3700+
Google Solution Challenge
2× $1,100 wins
Rapid Rebuild · LifeQuest
bmw challenge
Amulate · Continuum.ai
02
amu · circuits, clubs, and real responsibility
at zakir husain college of engineering & technology, i stopped treating engineering as coursework and started treating it as infrastructure—for myself and for hundreds of others on campus.
as web & tech lead at google developer groups on campus (ZHCET), i led ~13 people across three full-stack builds, mentored volunteers, and took our team's VoltSense (same track that's headed toward an ieee-venue paper on smart energy / anomaly detection) to the top 105 of 3,700+ in the google solution challenge 2025. later i stayed on as a mentor.
with AMURoboclub as joint coordinator, i mentored teams for TechnoXian 9.0, wrote flight and ground software for an IN-SPACe CANSAT mission (ASI), and shipped a Next.js + PocketBase registration stack for Vercera 4.0 (200+ signups).
IEEE — head of web operations, Code-o-Fiesta 3.0, technical org for AMUROVc 3.0.
public graph
code, issues, and experiments live on github—most of my portfolio is there before it's here.
builds in motion
the unpolished parts—solder smoke, failed runs, the one commit that finally greened—usually end up as reels on instagram before they end up here.
03
things i've shipped (the short list)
stick — react native app (play beta), express backend end to end, docker on AWS EC2, ci/cd, marketing site, keeping prod boring.
continuum — agentic workflows over slack: MCP, FastAPI, postgres, redis, celery, context across jira, github, calendar.
commitlens — open-source pr review agent: webhooks, detectors, learning from human comments, claude summaries, slack.
neurosense — wearable screening; esp32, fft, mediapipe + random forest, sub-100ms websockets. SIH 2025 national finalist.
zenventures — lifequest balance winner; next.js, firebase, gemini—with shashank chauhan.
more builds on the full work page.
motion
a reel, not a pitch
same piece as the home hero—mute by default; use the floating control if you want sound.
how i think about building
systems over features. a well-designed system naturally produces good features. spend time on architecture, not decoration.
constraints breed creativity. the best work happens when you have limits. work within them, and you'll find elegant solutions.
ship early, iterate always. perfect is the enemy of done. get something working, learn, improve.
tools i reach for
fine print
allergic to dust and pollen; breathing stuff that means i basically wake up and sleep with a cold most days. i compensate with probably too much coffee and diet coke. if i seem low-energy in a dusty room, it's not you—it's my lungs filing a complaint.