yo9esh
Yogesh, drawn in manga screentone

YOGESH KUMAR

Full-stack engineer who designs how the pieces connect.

type to start wpm 0 acc 100%

tab restart 1 projects 2 story 9 lo-fi 0 clicks on ? all keys

Chatterbox TTS pipeline · 2026

From a 48 GB server GPU to an 8 GB gaming card

Automated voiceovers for 2,655+ real-estate listings. Listing text is cleaned into speakable text, then voiced by Chatterbox, whose memory grew with input length until it needed a 48 GB RTX A6000.

VRAM, any input length
16–80 → 5 GB
infra cost
−87%
listings, zero OOM
2,655+

Problem

VRAM use was 16 GB at baseline and grew to 48–80 GB on long inputs, causing OOM crashes. Position overflows at token boundaries caused voice hallucinations.

What I did

Transformed S3Gen into bounded-memory chunked execution and patched the position overflows and token boundaries, so VRAM stays at ~5 GB at any length. Upstream, a Node.js sanitizer turns each listing into speakable text (house numbers, dimensions, currency, acronyms) capped at ~150 words.

PythonPyTorchCUDARunPodNode.jsMySQLBunnyCDN
Architecture

SystemStar Infotech · 2025–26

A 360° player that starts fast

A 360° video web player and the video pipeline that feeds it, built at SystemStar.

360° source footage
4K
HLS segments (vs 4 s)
2 s
720p → 4K ladder
ABR

Problem

Generic CDN defaults use 4-second segments, which slows time-to-first-byte and playback start for heavy 4K footage.

What I did

FFmpeg probes each source, picks the fastest available encoder (GPU when present) and transcodes an adaptive ladder from 720p up to 4K into HLS with 2-second segments. A keyframe is forced every 2 seconds so every rendition switches on the same boundaries, and a master playlist feeds the React + Three.js player.

ReactThree.jsFFmpegHLSPythonCDN
Architecture

CoralTeams · Coral hackathon · 2026

A 4 GB AI stack from a potato laptop

CoralTeams is a self-hostable Teams interface I built for Coral, which connects tools like GitHub, Jira and Slack. It ships with a Docker image. The stack needed a 1 GB+ Rust engine and a 3 GB local LLM, and my laptop couldn’t run either.

per month
$0
free platforms, 2 tunnels
4
stack, off-laptop
4 GB+

Problem

No high-spec laptop, and cloud free tiers kept rejecting card verification. The stack couldn’t even be tested locally.

What I did

The FastAPI backend runs in GitHub Codespaces (reached over SSH with the GitHub CLI), the LLM runs on Google Colab’s GPU, and ngrok tunnels expose both. The frontend sits on Vercel and the semantic engine with Coral on Render.

Tradeoff

Colab sessions expire and tunnel URLs rotate, so it’s for building and testing. Once everything works, it moves to a persistent server.

Architecture
4 more projects Show less Cursor for n8n · AgentSFX · AlchEmaiLyst · Schedulo

1st prize · Ideatex, GDG on Campus HIT · 2025

Cursor for n8n

Describe an automation in plain English, get an n8n workflow. A Chrome extension built in five days with Soumyadeep Bhowmik.

of 400+ developers
First
idea to pitch
5 days
prototype over n8n nodes
RAG

Problem

LLMs hallucinate n8n nodes and parameters, so AI-generated workflows can’t be trusted.

What I did

Ground the model before it writes: the extension fetches the relevant n8n docs through Context7, adds them to the prompt for the chosen LLM, and returns workflow JSON ready to import. A RAG prototype went further, searching n8n’s node database and workflow templates (FAISS) with Gemini.

JavaScriptChrome extensionPythonRAGFAISSSQLiteGemini
Code
Architecture

Track winner · Calcutta Hacks · Dec 2025

One-click game audio for Godot

Searching for and creating sound effects is a tedious part of making a game. We built a Godot add-on that creates SFX, character voices, NPC dialogue and BGM with AI and integrates them in one click. Team 4bajesonewale, with Soumyadeep Bhowmik, Harsh Sinha and Amlan Roy.

Calcutta Hacks
Track winner
generate + integrate
1 click
person team
4

Problem

Searching for and creating sound effects for a game is tedious.

What I did

An add-on inside the Godot editor: an LLM analyses the game’s code and a screenshot to list the sounds it needs, ElevenLabs generates them, and the add-on integrates everything into the code in one click.

Architecture

Open source · GSSoC ’25 Project Admin

AlchEmaiLyst, and the community around it

An email assistant with three agents: Summarizer, Composer and Spam. Then I opened it up through GirlScript Summer of Code and ran it as a project.

GitHub stars
42★
contributors onboarded
30+
Alchemyst leaderboard
#1

Problem

Help people get through their inbox: summarise mail, draft and send replies, and clear spam from one screen.

What I did

A React app that signs in with Google OAuth 2.0, reads and sends mail through the Gmail API, and uses Alchemyst AI for summaries and drafts, falling back to Gemini.

Architecture

Schedulo · Slack bot · Jan 2026

A Slack scheduler with local AI

Most companies won’t let internal data touch ChatGPT or Gemini. So I built a Slack bot that schedules meetings right inside Slack, with the AI running fully locally on Ollama.

Llama 3.2 on Ollama
Local
cloud LLM calls
0
+ config to set up
3 files

Problem

Internal company data can’t be sent to cloud LLMs like ChatGPT or Gemini.

What I did

Slack talks to a Python agent over Socket Mode. A local Llama 3.2 on Ollama reads each mention and returns the intent, a summary and the start time as JSON. A Google Apps Script web app creates the Calendar event, and the bot confirms in Slack.

PythonOllamaSlackGoogle Apps Script
Code
Architecture

  1. Class 9

    Lockdown Boredom

    Started coding because the lockdown was boring. Quit after two months: too hard, no reason to keep going.

    Jan 2022

    The Smartwatch

    78th of 400 at a GeeksforGeeks coding tournament. Won a smartwatch. Kept winning school contests, still built nothing.

  2. Dec 2024

    The Finale

    First semester. Some luck, mostly groundwork: our team made the Smart India Hackathon Grand Finale. We lost, and I found a purpose: to win, and to build.

    Yogesh at the Smart India Hackathon 2024 Grand Finale

  3. Jul 2025

    First Win

    1st prize at Ideatex with Cursor for n8n. Then Project Admin for AlchEmaiLyst in GSSoC.

    Dec 2025

    Ship Before the Interview

    Got an SDE internship at SystemStar by building a mock before the interview. Read it →

  4. Music

    Off the Keyboard

    I sing, on stage whenever I get the chance.

    Yogesh singing on stage, mic in hand

  5. Now

    For the Love of Building

    20+ hackathons later I stopped caring about winning.

    “Building is the one thing I never get bored of.”

TO BE CONTINUED

Want to build something together? Email me.

fbyogesh111@gmail.com
Yogesh smiling in a yellow kurta at an old iron gate
つづく