File PG/2026
Hero01 / 05
Ref. PG—2026
Opened Nov 2008
Student · Builder · Saharanpur, India

Parth
Goyal

(seventeen, if that matters)

Systems over symptoms.

I build tools that find the pattern before the outcome does — a dopamine loop before it becomes a habit, a grade before it becomes a regret. Built across late nights, exam pressure, and too much curiosity.

See the work ↓ Read the statement →
Age 17
Saharanpur, India
Status — building SYNAPSE
About

Not a senior
engineer.
Just the one
who shipped first.

EXHIBIT A
Photo pending

"Two years of building. Two products live. First 100 users, zero rupees spent finding them."

I'm Parth, 17, from Saharanpur. I build tools, think in systems, and ship things that probably shouldn't exist yet for someone my age.

It started with a simple question: why do hidden patterns predict outcomes better than effort does? In academics. In behavior. Eventually, in code.

I'm not pretending to be a senior engineer. I'm a student who moves faster than the syllabus, learns by building in public, and finds leverage where everyone else sees noise.
StatusStudent, Saharanpur, India
StackJavaScript / TypeScript, Python, React
FocusSystems that predict + systems that recover
ModeSelf-taught. Ships in public.
Work

Selected work

Three projects shipped
2024 — 2026
STATUS LIVE 2024
001 — Featured project
ClassPredictor (actually worked)

"Students don't fail because Class 11 is hard. They fail because the gaps compound silently, long before the report card shows it."

Most academic tools tell you what score you got. This one tells you where you're heading — before it's too late to change it.

The insight was simple: model subjects as a dependency graph instead of isolated marks — prediction beats motivation once a student can see the trajectory of their class division in real time.

Built under real exam pressure. Tested by real students. Shipped alone.
↗ Visit live project
100+Organic users
2 daysTo first 100 users
₹0Spent on marketing
Prediction logic
Input — subject-wise marks
Map — prerequisite dependency graph
Weight — risk score per node
Output — division + early warning
Built with
React Vite Firebase Firestore Dependency graph algo Vercel
STATUS BUILDING SEPT 2026
002 — In development
SYNAPSE (the hard one)

"You don't lack willpower. You have a hijacked reward system. This fixes it."

Dopamine addiction isn't a character flaw — it's an architecture problem. SYNAPSE maps your specific compulsion loop, then deploys a personalized AI protocol built to match how your particular loop broke.

A five-step confession flow extracts the pattern; streak logic, relapse response, and milestone drops keep it evolving. Every request runs through a server that verifies who you actually are — no client can fake its way in.

The hardest feature to build was the one that had to feel human.
↗ Preview at synapserewire.site
5Step confession
70BLLaMA model
∞Streak resets
Recovery protocol
Confess — 5-step addiction mapping
Analyse — AI pattern extraction
Battle plan — LLaMA 3.3 70B response
Track — daily check-ins + streak
Recover — relapse SOS + milestones
Built with
React Firebase Auth · Firestore · FCM Groq / LLaMA 3.3 70B Vercel Serverless PWA
STATUS LIVE 2026
003 — Independent build
Verdikt

"NSE moves in seconds. News takes minutes. Verdikt closes that gap."

An AI-powered market intelligence terminal for Indian markets — NSE, BSE, MCX, RBI. It ingests live RSS feeds from Moneycontrol, LiveMint, Economic Times, Hindu BusinessLine, and Zee Business, then runs every article through a Gemini-powered multi-agent evaluation pipeline for market-impact scoring and categorization.

Automatic API key rotation keeps the pipeline running with zero downtime, SHA-256 hashing deduplicates stories across sources in real time, and the terminal reads in English, Hindi, and Hinglish.

Frontend on Vercel, ingestion backend on Railway — built to process a live feed continuously, not just on request.
↗ Visit live project
5News sources ingested
3Languages served
0Downtime, key rotation
Evaluation pipeline
Ingest — live RSS from 5 market sources
Dedupe — SHA-256 hashing in real time
Evaluate — Gemini multi-agent impact scoring
Serve — EN / HI / Hinglish terminal
Built with
React Vite Node.js Supabase Gemini API Tailwind · Framer Motion
Method

How I think

Note A
※
Systems over hustle
I don't grind harder — I build things that keep working while I sleep. Every project starts with the same question: where's the hidden leverage point?
Note B
※
Patterns before people notice
Most people react to outcomes. I'm interested in the architecture underneath them — a knowledge gap, a broken reward loop, whatever's actually driving the result.
Note C
※
Ship early, learn publicly
The fastest way to learn is to build in public and let reality correct you faster than any course would.
— End of file —
Contact

Let's build
something
real.

Open to collaborators, hard problems, and people who think in systems. I answer everything.

Parthgoyal379@gmail.com
github.com/parthgoyal379-oss
— Parth
CASE OPEN
Subject dossier ✓ Verified
NAME — Parth Goyal
BASE — Saharanpur, India
BUILDING — SYNAPSE, shipping Sept 2026
INTERESTED IN — leverage, prediction, systems that scale
OPEN TO — hard problems, real collaborators