GitHubLinkedInInstagramResume
Tharun Rathod

TharunRathod

Engineer. Founder. Researcher.

Ten-plus production systems built solo and five research papers, alongside a B.Tech in Electrical Engineering at IIT Roorkee.

View flagship workResearchGitHub ↗Contact
About Tharun

Solo founder building products from 0→1. Ships production systems end-to-end - product vision, technical architecture, user acquisition, revenue. Edcore (300+ users, 4 live modules), Rune (optimizing compiler), SmartAsset (enterprise infrastructure), and 7+ shipped projects. Founder-engineer who writes code, talks to users, and owns P&L.

Education · Healthcare · Hiring · Developer Tools · Enterprise Systems · Research

Production systems
10+
Research papers
5
Empirical trials
484
Colleges dataset
2,000+
Code on screen

My B.Tech is in Electrical Engineering, but the real work is building companies. I'm a solo founder who ships production systems - from zero to live users - across education (Edcore, 300+ students), healthcare (PHC queue systems), developer tools (MCP servers, repo visualization), and enterprise infrastructure. Product strategy, full-stack execution, and go-to-market - all solo. The degree taught me circuits. The startups taught me everything else: customer discovery, unit economics, retention loops, and shipping fast under constraints.

Flagship

Flagship
Projects

Four production-grade systems built solo - from education OS to optimizing compilers.

Flagship

Edcore

All-in-one education OS for Indian students - live with 300+ users across exam prep, college selection, and career guidance.

edcore.tech

Four production modules under one unified Next.js 16 monolith: ExamNotifi (exam calendar + notifications), CollegeTracker (personalized college shortlisting), Tutorix (mentor matching), and NextTalk (career counseling booking). Shared authentication via NextAuth, unified CRM with Airtable sync, MSG91 OTP flows, demo scheduling, and email automation with Resend. Not four isolated apps - one cohesive platform with cross-module state management, MongoDB schema shared across all features, and a notification engine that tracks user behavior across modules.

300+
Active Users
4
Live Modules
2k+
Colleges
25
Data Fields
Education
SaaS
Next.js 16
TypeScript
MongoDB
NextAuth
MSG91

Engineering Insight

The real moat isn't the code - it's the 2,000+ college dataset curated across 25 fields (hostel vibe, coding culture, campus life, placements, city tier, faculty, infrastructure, location advantages). Competitors have college names. Edcore has the depth students actually need to decide. The dataset took months of manual curation and is harder to replicate than the entire platform.

Flagship

Rune

An optimizing compiler that infers the optimal algorithm from your intent - and formally verifies correctness.

Try in BrowserGitHub

Write what you want, not how to compute it. Rune parses intent-driven keywords (group, order, take, explore, maximize) and automatically selects the right algorithm: heap vs sort for TOP_K, BFS vs Dijkstra for shortest paths, Kadane's algorithm for maximum subarray. Three execution backends (interpreted Python, Numba JIT, native C++ with LLVM) with automatic backend selection based on problem size. Every optimization is verified against a naive baseline using property-based testing - 115 test cases ensure correctness across all optimization paths.

3
Backends
115
Tests
5x
Speedup
Verified
Correct
Compiler
DSL
Python
Numba
C++
LLVM-JIT
Algorithm Selection

Engineering Insight

The innovation isn't one clever optimization - it's a repeatable framework for algorithmic selection. TOP_K selection happens at parse time (syntax-driven). BFS vs Dijkstra happens at runtime (graph structure inspection). Kadane detection happens from objective analysis (maximize + contiguous). Three structurally different decisions across three algorithm families, all automatically verified. It's proof that intent-level programming can outperform hand-coded solutions while guaranteeing correctness.

Flagship

IITR Nexus

Unified campus intelligence - one query routes across academics, library, dining, events, and notices via parallel MCP retrieval.

GitHub

Campus information lives across 10+ disconnected systems. IITR Nexus routes a single query across academics, library, dining, events, and notices using multiple MCP servers in parallel. Intent classification determines which servers to query. Confidence scoring ensures grounded responses - if the system is uncertain, it surfaces the source, not a guess. Reasoning panel visible to users.

5+
Data sources
MCP
Architecture
Parallel
Retrieval
Grounded
Responses
Education
AI
MCP
Next.js
TypeScript
Groq Llama 3.3 70B
Intent Detection

Engineering Insight

Intent classification before routing reduces wrong-server queries significantly. The hallucination resistance mechanism is the key design choice - when confidence drops below threshold, the system returns a source reference instead of synthesizing an answer, preserving trust at the cost of convenience.

Flagship

Canopy

Repo dependency visualization - parses Python and JS/TS imports via the GitHub API. No cloning. No local setup.

Live AppGitHub

Canopy parses Python and JS/TS import statements directly from GitHub - no cloning, no local execution, no CI setup. Builds an interactive React Flow dependency graph with Gemini-generated module summaries. MD5 caching prevents redundant parses. Dark forest UI for extended analysis sessions.

2
Languages
0
Clones
MD5
Cache
AI
Summaries
Developer Tools
Next.js
React Flow
Gemini API
GitHub API
TypeScript

Engineering Insight

Parsing imports without executing code requires careful alias resolution - TypeScript path aliases, Python relative imports, and dynamic requires all break naive string matching. The parser handles edge cases before graph construction, because a dependency graph with silent resolution failures is worse than none.

AI & Machine Learning

AI & ML

AI/ML projects under real constraints - genuine AI where it earns its place.

IHateApplying↓

Bulk outreach automation for job applications - upload CSV, AI generates personalized emails/DMs, push to Gmail. Apply till you get a reply.

Cold email automation tool that generates personalized outreach emails and LinkedIn/X DMs at scale. Upload a CSV with contacts (email required, name/company/context optional), paste your resume and skills, and the AI generates customized drafts based on your background. Batch processing (50 rows at a time with free keys), Gmail integration to push drafts directly to inbox, auto-detects email column from any header format. Built for job seekers tired of manual applications.

Click to expand →

PHC Queue Management System↓

A token + wait-time system for Primary Health Centres, piloted against Latha Children's Clinic in Attapur, Hyderabad. Front desk issues a printed token with an honest wait estimate off the rolling 10-consultation average; doctor clicks through patients to advance the queue; front desk gets a phone-number callback list for no-shows. No app install for patients, no internet required for the queue logic itself, and V1 ships with zero third-party API dependency on purpose - every piece justified by an actual cost or reliability constraint.

Click to expand →

NIFTY-50 Investment Intelligence↓

Investment intelligence platform on 21 years of NIFTY-50 data. Hidden Markov Models detect bull/bear regime transitions. XGBoost price prediction with SHAP explainability. Portfolio optimization via Modern Portfolio Theory. Risk analytics with Value-at-Risk and drawdown analysis. Interactive Plotly dashboard.

Click to expand →

Developer Tools

Dev
Tools

Tools built for developers.

SmartAsset↓

Enterprise asset management - 56+ assets, zero double-bookings, full audit trail.

Replaces paper logbooks for warehouse-scale asset tracking. Date-window availability validation ensures no two reservations overlap. Atomic inventory updates and serializable approval logic prevent race conditions under concurrent booking. QR-based issue and return creates a frictionless audit trail. Analytics dashboard, role-based access control, and full warehouse workflow management.

Click to expand →

TravelChecker↓

Multi-modal travel comparison for India - MCP server + REST API + dashboard. Finds optimal composite routes like cab→airport→flight.

Built as a Model Context Protocol (MCP) server so it can be plugged into Claude Desktop, Cursor, or any MCP-compatible AI agent as a tool. Also ships with a REST API and a web dashboard. Most travel apps only show direct routes. Search "Khammam to Delhi" and you'll see slow trains - but never the fact that the fastest option is actually a cab to Vijayawada airport + a flight from there. TravelChecker finds the nearest transport hub (airport/railway station) for any city and constructs complete multi-leg journeys, then ranks them all by your priorities.

Click to expand →

Wayfound↓

AI-powered itinerary planner for Hyderabad. Tell it your group size, age, vibe and budget - it plans your whole outing with real venues, timings and cost breakdown.

Stop searching. Start going. Wayfound generates complete day plans for Hyderabad in seconds. Input your group (size, age), activity preference (go-karting, romantic dinner, gaming zone, cultural day), and budget - the AI outputs a full itinerary with venue names, addresses, timings, and cost breakdown. Uses Groq AI (llama-3.1-8b-instant) for natural language itinerary generation and Google Places API for real venue data. Smart mock engine with 200+ verified Hyderabad venues as fallback when API quota runs out.

Click to expand →

SurgeScope↓

Surge pricing analytics using arc elasticity on 50,000 Mumbai/Bangalore/Delhi trip records. Found the 2.0x wall where completion falls off a cliff.

We had 50,000 Uber/Ola-style trip records from Mumbai, Bangalore, and Delhi over a year and wanted to know: does surge pricing actually work, or does it just chase riders away past a certain point? We used arc elasticity, comparing completion rate (did the ride happen or get cancelled) at each surge step, from 1.0x up to 2.5x. That gave us a demand curve instead of a guess.

Click to expand →

Product Studies

Product
Studies

Real apps analyzed - case studies, competitive battles, and product proposals. Every UX and business decision broken down.

MAGGI CUSTOMS - Product Proposal for Nestlé India↓

A complete product hypothesis for Nestlé India proposing a modular instant-noodle ecosystem (MAGGI CUSTOMS) targeting Gen Z college students. 40-page document covering market opportunity, product architecture (MAGGI DUO, Flavor Sachets, Bowl format, Campus Bowl Lab), validation plan, retailer economics, and commercial case. Includes consumer hypotheses, campus pilot methodology, metrics framework, and illustrative scenario modeling.

Click to expand →

Tata Neu Product Teardown: Gen Z Engagement & Retention↓

Product Teardown 2026 case study diagnosing Tata Neu's "Buy and Bye" problem - students use the app for big purchases, then it goes dormant for months. Proposed solution: shift from transaction-led to relationship-led engagement through two integrated product cases that give students a reason to open the app without buying.

Click to expand →

Edit Journey - Product Proposal, Hyderabad Metro↓

Found a usability gap in Hyderabad Metro's QR ticketing: selecting the wrong destination forces riders into a multi-day cancel-and-refund cycle, with no way to just fix the fare. Designed and prototyped "Edit Journey" in Figma - an instant, cancellation-free fare-adjustment flow - then pitched it directly to Hyderabad Metro with a written product brief and outreach email. No one asked for this one; I noticed the gap and went to the source.

Click to expand →

⚔️ UX Battle

Food Delivery Battle - Swiggy vs Zomato vs EatSure vs Domino's vs KFC↓

Swiggy · Zomato · EatSure · Domino's · KFC

Found the framing effect, emotional onboarding, progressive disclosure, the say-do gap - and a deliberate anchoring exploit in Zomato's Food Rescue screen that inflated perceived savings by ₹70.

Click to expand →

Expertise

Skills

Tech Stack

🐍
Python
⚡
JavaScript
TS
TypeScript
⚛️
React
▲
Next.js
🟢
Node.js
🍃
MongoDB
🐘
PostgreSQL
C++
C++
🤖
Python ML
📦
Git
🐳
Docker
Full-Stack Development
System Architecture
API Design & Integration
Database Design & Optimization
AI & LLM Systems
MCP Server Development
Compiler Design
Algorithm Optimization
DevOps & Cloud Deployment
Product Management
Business Analysis
UX Research & Testing
Technical Writing
Agile & Scrum
Research & Experimentation
Next.js 16
React
TypeScript
JavaScript
Tailwind CSS
Framer Motion
shadcn/ui
React Flow
Vite
Responsive Design
Career

Work
Experience

Founder and Product Engineer

Edcore

Jan 2025 - Present

Built and launched an education platform helping Indian students discover colleges, exams, mentors, and career pathways. Grew the platform to 300+ users and generated 2.7K+ impressions through organic distribution and experimentation. Took Edcore from problem discovery → UX → full-stack development → deployment → user acquisition.

Next.js
PostgreSQL
NextAuth
Product Ownership
User Acquisition

Independent Researcher

IIT Roorkee

2025 - Present

Five papers across AI evaluation, behavioral AI, electrical engineering and linguistics - two sole-authored, two co-authored, one published open-access on Zenodo. Current focus: whether a model can be trusted about its own output.

LLM-as-Judge
Experimental Design
Academic Writing

Forward Deployed Engineer

R.K. IntelliServe (Startup)

Mar 2025 - Mar 2026

Pipeline Orchestration: Architected and automated DAG-based ML preprocessing and model evaluation pipelines for high-throughput production data workflows. Client and Systems Integration: Translated client requirements into scalable AI/ML solutions, integrating data systems, APIs, and existing enterprise workflows. Full-Stack Engineering: Built low-latency microservices, third-party API integrations, and automation scripts to streamline operational workflows for SME clients. Production Engineering: Collaborated across engineering pods to troubleshoot integration failures, ship emergency hotfixes, and stabilize production systems under tight delivery timelines.

Python
ML Pipelines
DAG Orchestration
Airflow
Startup
Client Integration

Web Development Intern

Hiring Bazaar AI

Oct 2025 - Jan 2026

Product Delivery: Shipped the careers page, role listings and candidate application workflow end to end (frontend and backend); owned UI feedback and rollout testing. Performance Optimization: Cut average API response time 30% (400 ms to 280 ms) and initial page load 25% (2.4 s to 1.8 s).

React
Full-Stack
API Design
Performance
TypeScript
Mean Stack
Papers & Publications

Research

Five papers spanning AI evaluation, behavioral AI, electrical engineering, and linguistics. Two sole-authored, two co-authored, one published open-access.

Background

Education

Indian Institute of Technology Roorkee

B.Tech, Electrical Engineering

Indian Institute of Technology Roorkee

All India Rank 23,496 in JEE Advanced 2023 and Telangana EAPCET Rank 2,614. The degree is in electrical engineering; most of the building happens outside the syllabus - ten production systems, five papers, and two products shipped solo alongside coursework.

2023 - 2027

Telangana State Board · CBSE

Class XII & Class X

Telangana State Board · CBSE

Class XII - 94.20% (Telangana State Board). Class X - 95.60% (CBSE). Also served as an NCC Cadet with the National Cadet Corps at IIT Roorkee between October 2023 and April 2024.

2021 - 2023

Get in Touch

Let's talk

Always open to opportunities - full-time roles, internships, freelance gigs, or just interesting conversations about tech, products, and building things that matter.

Whether you're hiring, have a technical challenge to discuss, want to collaborate on a project, or just want to connect - I'd love to hear from you. Drop me an email or find me on the platforms below.

Resumes

Resume 1
↗

Software Development

Full-Stack Engineering & Product

Download PDF→
Resume 2
↗

AI Research

Papers & LLM Systems

Download PDF→
Resume 3
↗

Product Management

Implementation Engineering & Strategy

Download PDF→

Or reach me directly

tharunrathod2005@gmail.comedcore.tech

© 2026 Tharun Rathod - Engineering, Products & Research

Roorkee / Hyderabad / Remote