Vedangi Thokal

Experience

Jun – Aug 2025

Bangalore, India

Software Engineering Intern

Intuit · QuickBooks Online

Built a plug-and-play TypeScript widget for a new QuickBooks Online feature, enabling reuse across product surfaces.
Collaborated on API design, aligning frontend and backend contracts and influencing the product's technical architecture.
Built a JIRA MCP server with Slack integration — reduced developer context-switching by ~25%.

Expected 2026

Pune, India

B.Tech — Computer Science Engineering

MIT World Peace University · AI & Data Science

Specialisation in Artificial Intelligence and Data Science.
Active across research, technical leadership, and social initiatives.

01 / 03

System Design Lab

Next.js · Typescript · PGLite · xterm.js · MDX

An interactive educational platform for system design — where you don't just read about databases, you run them live in your browser tab.

Zero-latency client-side execution engine: full PostgreSQL compiled to WASM, running entirely in RAM.
Simulates concurrent transactions, isolation levels, and anomalies without a single backend request.
Serverless architecture — no infra costs, no cold starts, instant for every user.

Visit the website →
Source →

 ┌──────────────────────────────────┐
 │  browser tab  (zero server)      │
 │                                  │
 │  $ SELECT * FROM accounts;       │
 │  > id  name    balance           │
 │  > 1   Alice   $1,200   ✓        │
 │                                  │
 │  $ BEGIN;                        │
 │  $ UPDATE accounts               │
 │      SET balance = balance - 100 │
 │      WHERE id = 1;               │
 │  $ COMMIT;  ✓                    │
 │                                  │
 │  engine:  PostgreSQL / WASM      │
 │  latency: ~0ms   no network      │
 └──────────────────────────────────┘

02 / 03

Devtrace

Node.js · PostgreSQL · Express

A pluggable backend observability framework that automatically surfaces what's slowing your application down — without any manual instrumentation.

Detects slow API routes, expensive SQL calls, N+1 query patterns, and memory spikes in real time.
Full-stack dashboard with a batched event-ingestion pipeline, backpressure handling, and sampling.
Designed as a reusable drop-in package — add one line and it works.

 ┌──────────────────────────────────┐
 │  observability / backend         │
 │                                  │
 │  GET /api/data  →  1842ms        │
 │  ████████████████████░░  SLOW    │
 │                                  │
 │  SELECT *  →  n+1 detected       │
 │  ████████░░░░░░░░░░░░  WARN      │
 │                                  │
 │  heap: 187mb ──► 340mb           │
 │  ██████████████████████  SPIKE   │
 │                                  │
 │  latency p99:  2100ms            │
 │  queries/req:  14   ← too many   │
 └──────────────────────────────────┘

03 / 03

Forest Protection System

Arduino · NodeMCU · Supabase · Next.js · Blynk · IoT

A real-time IoT system that monitors remote forest edges for fire risk and illegal activity — combining hardware sensors with a cloud-connected web dashboard.

Temperature, humidity, and tilt sensors detect fire conditions and illegal tree-cutting in real time.
RFID access control logs and restricts entry to restricted zones.
Wi-Fi via NodeMCU (ESP8266) streams data to Supabase; Blynk pushes instant mobile alerts.

 ┌─────────────────────────────────┐
 │  forest edge · sensor node      │
 │                                 │
 │  temp:      38.4°C   ▲  HIGH    │
 │  humidity:  11%       ▼  LOW    │
 │  tilt:      MOTION DETECTED !!  │
 │  RFID:      UNAUTHORISED  !!!   │
 │                                 │
 │  ESP8266 ──► WiFi ──► Supabase  │
 │  Supabase ──► Next.js dashboard │
 │  Blynk    ──► mobile alert SENT │
 │                                 │
 │  uptime:   99.2%                │
 │  alerts sent today:  3          │
 └─────────────────────────────────┘

Research

  input: leaf image (corn)
          │
          ▼
  ┌────────────────────┐
  │   Conv2D + ReLU    │
  │   ▓▓░░▓░▓▓░░▓▓░░   │
  │   feature maps ──► │
  └─────────┬──────────┘
            │ MaxPool
            ▼
  ┌────────────────────┐
  │   Conv2D + ReLU    │
  │   ░▓▓░▓▓▓░░▓░▓░    │
  └─────────┬──────────┘
            │ Flatten
            ▼
  ┌────────────────────┐
  │   Dense → Softmax  │
  └─────────┬──────────┘
            │
     ┌──────┴───────┐
     ▼              ▼
  healthy        diseased
  p = 0.02      p = 0.98

Plant Disease Detection using Deep Convolutional Neural Networks for Corn, Rice & Banana Crops

Built a deep CNN model that classifies crop disease from leaf images — making visible the patterns of illness that escape the human eye. Trained and evaluated across three crop types with high classification accuracy.

3rd ICISA 2024 · MIT-WPU, Pune, India

Leadership

Jun 2024 – Jun 2025

President

Computer Society of India, MIT-WPU

Oversaw 27+ club activities — project facilitation, student mentorship, and city-wide technical events. Led a deep learning research group and a DSA focus group of 12+ members.

Apr – May 2024

Section Leader

Stanford Code In Place 2024

Selected from 1,000+ applicants to guide a global cohort of 10+ students through Python. Fostered collaboration across diverse backgrounds and time zones.