I’m a software developer from India. I finished a Bachelor of Computer Applications at SRM Institute of Science & Technology in June 2026 with a CGPA of 9.42, and I’m now on the Master of Computer Applications there.
The work I have enjoyed most tends to arrive after a feature already works. On ChronoFlow that meant tearing up an indexing scheme that looked fine right up until the schedule views went under load. On the sentiment pipeline it meant working out what a dashboard should honestly show when a stage is degraded and the numbers are only half there.
Open source
I contribute to GRASS, a geospatial analysis system with a long history and a lot of users who depend on it behaving exactly as it did last year. My largest piece of work there brought space-time datasets into the Data Catalog GUI, which had only ever queried the spatial database, so anyone working with time series was stuck driving them from a terminal. It is merged.
Most of what I took from it was not the feature. It was working inside a codebase far older than anything I had touched, having design decisions argued down by people who knew the system better, and learning that the honest move when a change outgrows your understanding is to stop and say so.
The engineering write-up is in the Data Catalog case study , and the week-by-week record is in my 2026 journal .
Selected work
ChronoFlow is a full-stack scheduling platform for meetings, deadlines and leave workflows. Spring Boot and a normalized MySQL schema underneath, a hand-written browser client on top, JWT and role-based access between them. Rebuilding its indexes took 30% off read latency at peak load.
A real-time YouTube sentiment pipeline on Google Cloud. Cloud Functions ingest, Pub/Sub absorbs the bursts, Dataflow classifies multilingual comments in under a minute end to end, and a Streamlit dashboard makes the result readable.
An ATS resume ranking engine that compares what a resume and a job description mean, using embeddings, so a candidate is not penalised for choosing different words than the posting did.
How I work
I measure before I optimise. The slow reads on ChronoFlow only showed themselves under peak-load testing. Until I ran it I was guessing at which queries mattered, and I would happily have spent a week tuning the wrong ones.
I decide what happens when things go wrong before I ship. Very little of the sentiment pipeline is the model. It is buffering, retry limits, and what a dashboard should show when one stage is degraded and the numbers are only half there.
I write things down while I still remember why. A change only I can operate is not finished, so documentation goes in with the code and not in a later pass that never comes.
Algorithms and problem solving
I keep this going in small regular sessions:
- LeetCode: 1000+ problems solved, contest rating above 1900.
- TCS CodeVita: global rank 786 in Round 1, top ~1% of over 100,000 participants.
It sharpens my sense of what an approach will cost before I commit to it. It is supporting evidence for the engineering, and no substitute for shipping any of it.
Certifications
Machine Learning Specialization and Deep Learning Specialization from DeepLearning.AI, plus the Advanced Data Analytics Professional Certificate and the Data Analytics Professional Certificate from Google.