I build real-timesystems for placeswhere downtimeisn't an option.
I'm Rohit Ghadi, a Software Engineer at Partronics. I build real-time, full-stack systems — from backends that hold under peak load to the interfaces people depend on — and I optimise for the hour everything is hardest, not the average one.
Rohit Laxman Ghadi
Software Engineer · Partronics
- Focus
- Real-time systems
- Domain
- Passenger info · Indian Railway
- Based
- Pune, India
- Status
- Open to work
A system is a set of decisions you can defend.
Four that shape everything I build. Each one has a number attached to it, because a principle that can't be measured is just a preference.
Reliability is a feature, not an afterthought
On a station kiosk, a crash is a passenger who misses a train. I hunt memory leaks and tune backends for peak-hour load — cutting crash rates 20–30% on Par Kiosk because uptime is the product.
Design for the load you'll have, not the load you have
Suburban traffic spikes are predictable if you respect them. I build data layers that hold their response times under peak — a 15–20% improvement in load handling came from designing for the worst hour, not the average one.
Own the interface to the real world
When an off-the-shelf feed didn't exist, I designed a dedicated Suburban API rather than bend the product around a gap. Systems that last own their boundaries instead of inheriting someone else's.
Measure, then move
Every change earns its place against a number — crash rate, load handling, devices recovered. Patterns over hunches; evidence over adjectives. It's the same discipline that kept a 9.58 / 10 CGPA.
Selected systems, on file.
A registry of what I've shipped — open one to read the detail. The stack always serves the outcome, never the other way around.
- PRJ.01Production · Partronics
Par Kiosk
Passenger-information kiosks live across Indian Railway stations — train status, PNR, charting and wayfinding on terminals that can't crash at rush hour.
- Node.js
- NTES API
- Suburban API
- Real-time
Case study on request20–30%fewer crashes
Capability, read as a stack.
Full-stack development, machine learning, SQL, and AI — not a wall of logos, but the layers I actually reason in, from the languages at the base to the focus areas that hold it together. Filled chips are where I'm strongest.
Languages
ordered by reach
06 entries
- PythonPandas · Matplotlib · pytest
- JavaScript
- TypeScript
- SQLMS SQL Server
- Java
- C
Frameworks & runtime
where products get built
04 entries
- React.js
- Node.js
- AngularJS
- PyTorchdeep learning
Systems & tooling
ship and operate
04 entries
- Docker
- Git
- Azure
- MS SQL Server
Focus areas
where I go deep
04 entries
- Full-stack developmentReact · Node
- Machine LearningPyTorch
- AI tooling
- Data analysisPandas
The track record, in order.
Where the principles met production. Short history, real numbers.
Partronics
Software Engineer
- Building Par Kiosk, a passenger information system live across Indian Railway stations.
- Integrated the NTES API and designed a dedicated Suburban API for real-time train details.
- Cut crash rates 20–30% by hunting memory leaks and tuning backends for peak-hour load.
Softlabs
Software Engineer Intern
- Improved load handling 15–20% by tuning backend modules for consistent peak-hour response.
- Diagnosed and recovered 13+ devices, restoring them to working condition.
Education
B.S. Information Technology
Sant Rawool Maharaj College · University of Mumbai
Certifications
- Machine LearningInternshala
- Microsoft SQL Server from ScratchUdemy