DEV Community: Pranav Gupta

Week 2 | Days 1–2 – Backend Foundation

Pranav Gupta — Sat, 22 Aug 2026 16:10:59 +0000

Progress on the Multi-Source Analytics prototype:

FastAPI app structure, health check, and local dev setup
SQLite + SQLAlchemy persistence layer
Core models: cases, file uploads, normalized records
Confirmed data writes and reads correctly from disk

Hit a classic missing-package error — I wasn’t inside the virtual environment. Activated the venv and everything worked.

This persistence layer is important because the upcoming upload and parse APIs need a reliable place to store data.

Next: connect the file parsers and start the upload flow.

Building in public. Feedback welcome.

fastapi #sqlalchemy #python #hackathon #buildinpublic

Week 1 Done – Foundation is Ready

Pranav Gupta — Mon, 17 Aug 2026 16:05:01 +0000

Hey everyone

Just finished Week 1 of the project.

Here’s everything that got completed:

Set up the full project structure
Generated a synthetic multi-source dataset (used Faker and tried to make it feel realistic with Indian names, patterns, etc.)
Built a multi-format file parser that handles CSV, JSON, and XLS/XLSX
Added a timestamp normalizer that converts everything to ISO-8601

The individual pieces were manageable, but connecting the parser and normalizer properly took more effort than expected. I had to decide the right order and clean up the flow so they work as one pipeline instead of two separate scripts.

Final flow right now:
raw folder → parse → normalize timestamps → clean output

Basic data ingestion is in place. Next focus is storing the data and starting to link relationships between records.

If you’ve built similar data pipelines, would love any suggestions.

Building this in public for the Chandigarh Police Hackathon.

Building a Police Link Analysis Tool in 4 Weeks (Hackathon Time!)

Pranav Gupta — Wed, 12 Aug 2026 17:02:28 +0000

Hey Dev community!

I’m jumping into a hackathon sprint with a 4-week deadline (final submission: Sept 8).

Here's the problem I'm tackling: When police try to connect the dots between suspects, bank statements, and burner phone logs, they're usually stuck manually digging through messy, disconnected spreadsheets.

I’m building an end-to-end tool to solve this—turning siloed data into an interactive visual web that automatically finds hidden links between entities.

Keeping the stack simple so I don't get stuck in setup hell: Python, SQLite, NetworkX, Streamlit, Pyvis, and Gemini API.

The Plan
I'll be posting raw weekly progress, UI snapshots, and the inevitable debugging headaches here.

No fluff—just pure building in public. Let's see if I can pull this off before Sept 8!

Drop a comment if you've built anything with NetworkX or Pyvis before—would love any tips!