Associate Software Engineer at Folium AI, building backend services and useful developer tools.
Working across Django, NestJS, PostgreSQL, Redis, Celery, and React/Next.js.
About Me I'm Muhammad Mohsin, a software engineer based in Lahore. At Folium AI, I build backend features for client products and contribute to React/Next.js interfaces for an internal platform. I also build tools that make engineering workflows easier to run and maintain. Work at Folium AI
- Backend services: Built Stripe subscription flows, Django Channels WebSocket notifications, and NestJS audio processing for client products.
- Client savings assistant: Built a LangGraph assistant that routes user intent to goal creation, updates, listing, and status tools so users can manage goals through chat.
- Employee review workflows: Worked on Django performance and notification services and React/Next.js review screens. The UI shows KPI criteria while keeping employee scores hidden until finalization. Featured Projects Aevora + Utils
- Problem: Separate local setups made developer utilities harder to share and update.
- Solution: Built authenticated Django/Next.js chat that selects relevant tools and sends PR review, status report, and workspace setup tasks to a shared Celery worker through Redis. Aevora streams chat and task events to the browser via SSE.
- Impact: Users can request these tools from one interface while their dependencies and updates stay in the shared worker. OneDrop
- Problem: Sharing a sensitive file through a hosted service creates another stored copy.
- Solution: Built a Python HTTPS sender for direct LAN/VPN transfer with capability links, token checks, SHA-256 integrity display, and token-redacted audit logs.
- Impact: A link permits one download by default, then expires and shuts down the server. LeanGuard
- Research focus: Exploring motorcycle blind-spot detection in CARLA with simulated radar and IMU inputs and lean-angle filtering.
🛠️ Technologies & Tools Backend
Frontend
AI & Research
Infrastructure & DevOps
Appreciate you stopping by. Always happy to connect, learn, and build meaningful things.
⭐️ Fact: AI is trained on human data. That’s why it estimates a three-month sprint, even though it can build the project in three minutes





