I don't just ship code — I build systems that work when it counts, for the people who need it most.
I'm a B.Tech CS & AI student at Newton School of Technology, building at the intersection of AI safety, developer tooling, and underserved communities. My current obsession: making AI verifiable, robust, and genuinely useful - not just impressive on a demo day.
- 🌾 AgriNexus — Multimodal LLM platform for Indian farmers. Prompt-engineered for verifiability & failure mode analysis.
- ⚖️ Indian Legal Assistant — RAG pipeline (Llama3 + ChromaDB) with intent classification & toxicity filtering.
- 🔧 Jenkins.io Contributor — 6+ merged PRs; improving open-source developer experience at scale.
|
The OS for Indian Agriculture Multimodal LLM platform for crop disease detection built for farmers with minimal digital literacy. Key focus: verifiable outputs and graceful failure handling — because a wrong answer here means a lost harvest.
|
Making Law Accessible RAG pipeline using Llama3 + ChromaDB that retrieves and reasons over Indian legal text. Features intent/sentiment classification and toxicity filtering — built with safety-first architecture from day one.
|
|
Open Source at Scale Active contributor to Jenkins.io — 10+ PRs improving contributor spotlights and website UX. Part of the GSoC 2026 proposal track for "Retooling Jenkins.io Contributors-Spotlight."
|
QR-Based Smart POS Full-stack QR-menu and management SaaS with Razorpay payment integration. Currently deployed and tested with real restaurant clients in New Delhi.
|
class Yatharth:
def __init__(self):
self.role = "CS & AI Student @ Newton School of Technology"
self.location = "Delhi-NCR"
self.obsession = "Top 1% in at least one domain"
@property
def now(self):
return {
"building" : "AgriNexus — AI for Indian farmers",
"researching" : "Scalable Oversight & AI Control",
"contributing": "Jenkins.io | GSoC 2026",
"learning" : ["LLM Safety", "MERN Stack", "Competitive Programming"],
}
@property
def long_game(self):
return "Build verifiable, trustworthy AI for communities that can't afford it to fail."
"The best time to build something that matters was yesterday. The second best time is right now."