Task 1 — “Court-Data Fetcher & Mini-Dashboard”
Objective
Build a small web app that lets a user choose a Case Type and Case
Number for a specific Indian court, then fetches and displays the case
metadata and latest orders/judgments.
Court to target (example)
• Delhi High Court (https://delhihighcourt.nic.in/) or
• Any single District-Court eCourts portal (e.g., Faridabad District
Court on https://districts.ecourts.gov.in/).
Pick whichever you can reliably scrape within time; declare the
choice in your README.
Functional requirements
1. UI – simple form (HTML or lightweight framework) with
dropdown/inputs: Case Type, Case Number, Filing Year.
2. Backend – on submission, programmatically request the public
court site, bypass any view-state tokens/CAPTCHA (document
your approach), and parse:
o Parties’ names
o Filing & next-hearing dates
o Order/judgment PDF links (show at least the most recent
one)
3. Storage – log each query & raw response in SQLite / PostgreSQL
(any RDBMS).
4. Display – render the parsed details nicely; allow downloading any
linked PDFs.
5. Error handling – user-friendly messages for invalid case numbers
or site downtime.
Deliverables
Item Details
Code repo Public GitHub with MIT or Apache-2.0 licence
Court chosen, setup steps, CAPTCHA strategy, sample
README
env vars
Item Details
Demo video ≤ 5 min screen-capture showing end-to-end flow
Optional Dockerfile, pagination for multiple orders, simple unit
extras tests, CI workflow
Tech freedom
Use any language or stack you love (Python/Flask, Node/Express, Go,
etc.). Feel free to leverage Playwright, Selenium, or headless-browser
AI helpers (Claude Code, GPT-4o) for navigation.
Evaluation hints (for you, the interviewer)
• Robustness against site layout changes
• Code clarity & security (no hard-coded secrets)
• Documentation quality
• UI/UX simplicity
• Creativity in circumventing CAPTCHA legally (e.g., remote
CAPTCHA-solving service, manual token field, or court API if
available)
Task 2 — “WhatsApp-Driven Google Drive Assistant (n8n
workflow)”
Objective
Create an n8n workflow that listens to WhatsApp messages and
performs Google Drive actions such as listing files, deleting, moving,
and summarising documents in a folder.
Functional requirements
1. Inbound channel
o Use Twilio Sandbox for WhatsApp (or any WhatsApp Cloud
API) as the entry point.
o Parse simple text commands, e.g.:
§ LIST /ProjectX — list files in /ProjectX
§ DELETE /ProjectX/report.pdf
§ MOVE /ProjectX/report.pdf /Archive
§ SUMMARY /ProjectX — return a bullet digest of each
file’s content (PDF/Docx/TXT).
2. Google Drive integration
o OAuth2 connection; operate only within the authenticated
user’s drive.
o Use MIME-type aware nodes for reading content.
3. AI summarisation
o Call OpenAI GPT-4o (or Claude) inside n8n to produce
concise summaries.
4. Logging & safety
o Maintain an audit spreadsheet/log.
o Guard against accidental mass deletion (e.g., require
confirmation keyword).
5. Deployment
o Provide a ready-to-import workflow.json plus instructions
for running n8n via Docker.
Deliverables
Item Details
GitHub repo Contains workflow JSON, env-sample, helper scripts
Setup of Twilio sandbox, Google credentials, command
README
syntax
≤ 5 min showing WhatsApp conversations triggering
Demo video
Drive actions
Optional Slash-command help, natural-language parser,
extras webhook-secured endpoint
Allowed tools
n8n nodes, JavaScript functions, external AI APIs, or even a small
middleware microservice if n8n alone can’t handle a step.
Evaluation hints
• Workflow clarity (use node groups & comments)
• Error handling & user feedback in WhatsApp
• Security of tokens and scopes
• Elegance of AI summary prompts
• Extensibility (easy to add new commands)
General Instructions for Candidates
Aspect Expectation
Deadline 10 August 2025, 23:59 IST
We’ll share a Google Form within 48–72 h. Submit:
Submission
repo URL, short description, and demo-video link
form
(YouTube or Loom).
Even if you can’t finish every feature, push working
Partial credit
code and document what’s left.
Work solo; you may use open-source libraries / AI
Collaboration
tools but cite them in the README.
If stuck, document blockers in the README rather
Support
than ghosting—communication counts.
Applicants who deliver both tasks (even at MVP level) will be
considered first, but a polished single-task submission can still win the
internship.
Good luck—build something cool!
