Software engineer, applied machine learning · Bangalore, India
I'm a software engineer in Bangalore. At TrieCoders I build retrieval pipelines and work on bringing machine learning into how we test. Before that I made a parking system that spots free bays with a camera instead of a sensor in each one, and a fraud detector that took first place at Barclays Hack-O-Hire.
I build software that works on the hardware people already have.
I did my B.Tech in Computer Science and Engineering at SRM Institute of Science and Technology, specialising in AI and machine learning. I graduated in 2025 with a CGPA of 9.40.
I care most about the people at the edge of what a system assumes: the small operator, the public hospital, someone checking whether a building has a working ramp. The places that need these systems most are often the ones that cannot pay for the servers and sensors the systems take for granted. A lot of what I build starts from that limit.
Next I want to own a model in production: its evaluation harness, and the argument for its architecture. Further out I would like to work a layer down, on inference infrastructure, and then start a company of my own.
Through all of it the aim stays the same. I want AI to take the repetitive, tedious parts out of ordinary days.
Now
Software Development Engineer II at TrieCoders
Based in
Bangalore, India
Studied
B.Tech CSE (AI and ML), SRMIST, CGPA 9.40
Work
Jun 2026 – present
Software Development Engineer II
TrieCoders, Bangalore
Building retrieval pipelines for a system that is split across many services.
Bringing machine learning into testing in our CI pipeline.
Helping plan the architecture.
Aug 2025 – Jun 2026
Graduate Executive Trainee
Fidelity Investments, Bangalore
Joined full time after my internships. The first three months were the company's training programme, in Chennai.
After that, cloud infrastructure, CI/CD and backend work.
Jan – Jun 2025
Software Development Engineer Intern
Fidelity Investments, Bangalore
Automated a Swagger task with a utility in Node.js, TypeScript and AngularJS. It gave the team back weeks of associate time.
Jun – Aug 2024
Software Development Engineer Intern
Fidelity Investments, Bangalore
Built a tool that generates GraphQL queries on the fly. It was used for customer personalisation and for training ML models.
Oct 2021 – Jan 2023
Technical Executive
Alexa Developers SRM, Chennai
Moved up to executive in my second year.
Built the event backend from nothing, and looked after the backend teams for the Alexa Skills and GraphQL events.
Jan – Jun 2022
Software Development Engineer Intern
TrieCoders, Bangalore
Worked on the interface, the APIs and the reporting layer of a hospital management system covering pharmacy, laboratory, HR and accounting.
Partway through I was sent on-site to connect a client's existing systems to ours. It was accepted without rework.
Later I helped win new business and worked with clients directly.
What I took from it My first pitch to a room of hospital administrators was a tour of how the modules were built. Nobody asked a single question. Afterwards a senior manager told me what they had wanted to know: what would change about their working day. I had answered a question nobody in the room was asking.
Projects
01
SPARK
Finding free parking spots with a camera, on hardware a small operator can afford
Illustration
2024 – 2025Paper at ICCCNet-2025
Computer vision
IoT
Edge hardware
Research paper
What I built
A car park usually needs a sensor in every bay to know which ones are free. SPARK uses a camera instead, and it runs on small, affordable hardware. A lot that could never justify a sensor per space can still show drivers where to go. The system tells drivers in real time which spaces are free, and a pricing engine adjusts what a space costs from live demand, so operators earn more while the lot is used well. It connects to public transport and ride-sharing services, there is a mobile app for drivers, and an admin dashboard gives operators and city planners something concrete to act on for traffic flow and resources. I worked on every one of these parts.
How
I tried several ways of detecting a free spot and pushed each one through changing light and cars hiding other cars, until the results stopped surprising me. Then I measured how far I could shrink the model before accuracy started to slip, so the trade-off came from numbers. It grew out of my Computer Vision course, first as a semester project and then as my major project. I co-authored the paper and presented it remotely at ICCCNet-2025, hosted by Manchester Metropolitan University.
What I took from it
I now treat the hardware budget as a requirement from day one. It changed which models were even worth trying.
02
Quant Hawk
Fraud detection built in 24 hours, first place at Barclays Hack-O-Hire 2024
Illustration
20241st place, Barclays Hack-O-Hire 2024
Anomaly detection
Isolation Forest
K-means
Hackathon
What I built
A fraud detector built in 24 hours. It combines an Isolation Forest with K-means clustering to flag suspicious transactions across customer classes. We were one of 27 teams picked from over 1,000, with more than 4,000 people taking part across Pune, Mumbai and Chennai, and we came first. Barclays offered me a position after the win.
How
Around hour fourteen it all went wrong. The models sat under the accuracy bar, they were overfitting, and none of us could say why they generalised so badly. So we stopped and took a proper break, then went back to basics. After that we split in two. One half kept repairing the models. The other half went after the founding idea itself, to check it held up, and rebuilt the pitch so that technical and non-technical listeners would both get something out of it. Both halves landed before the 24 hours ran out.
What I took from it
Pushing harder on the same code rarely fixed anything. What got us moving was letting half the team question the premise.
03
Deshatan
A community bulletin for accessible public places
Illustration
2023Silver Medal, SRM Research Day 2023
Flutter
Firebase
Accessibility
Civic tech
What I built
An app for the question you ask before you leave home: will this place work for me? Residents flag barriers such as a broken ramp or an inaccessible entrance at public landmarks, and people with access needs can check before they go.
How
My team and I built it in Flutter on Google Firestore and Firebase Realtime Database. The reports come straight from residents.
04
Replete
Routing restaurants' edible surplus to people who need it
Illustration
2022Top 10, Mozohack 3.0
React
Node.js
Express
Supabase
PostgreSQL
What I built
A platform that lets restaurants send their edible surplus to people who need it. We built it for Mozohack 3.0, the hackathon run by SRMKZILLA, the Mozilla campus club at SRM, and it was recognised as an outstanding community and environmental project.
How
React on the front end, Node and Express over Supabase and PostgreSQL behind it. We redesigned the idea several times before one of them held together.
05
Asgn
Making student workloads visible across courses
Illustration
IEEE Student Branch, SRMIST
Regression
Scheduling
Product thinking
What I built
A tool that shows how much coursework is landing across all of a college's courses, so that deadlines stop piling up in places nobody is watching.
How
It started as an experiment during my time as a Research and Development Associate with the IEEE Student Branch: could a regression model spread assignments evenly across courses? It grew into a tool for tracking and transparency.
What I took from it
The model was the easy part. The gap was that nobody could see the workload building up.
06
Event backend
Registration, RSVPs and QR attendance for a student developer community
Illustration
2021 – 2023Used live at events
Express
TypeScript
Supabase
PostgreSQL
AWS SES
AWS S3
What I built
The backend behind the events of Alexa Developers SRM, written from scratch. It handles registration, sends RSVP emails automatically and checks people in by QR code.
How
Express and TypeScript over Supabase and PostgreSQL, with AWS SES for the email and S3 for storage, deployed on Heroku and Azure. It ran at live events.
07
Driver assistance
An upcoming project with vehicle manufacturers and OEMs
Illustration
Coming soonReserved
Computer vision
Automotive
OEMs
What it is
An upcoming driver-assistance project, with vehicle manufacturers and OEMs. I will share more when I can.
The details
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris.
Off limits for now. The lorem ipsum is on purpose.
Education
SRM Institute of Science and Technology
B.Tech, Computer Science and Engineering, specialisation in AI and Machine Learning
Kattankulathur, ChennaiSep 2021 – Aug 2025CGPA 9.40 / 10
The Amaatra Academy
CBSE Class XII, Science
Bangalore90% overall, 95% in Computer Science
Bishop Cotton Boys' School
ICSE Class X
Bangalore91.4% overall, 99% in Computer Science
The two courses I enjoyed most
Formal Language and Automata
Once I saw that a lexer is a finite automaton and a parser is a pushdown automaton, Compiler Design stopped being a list of stages and became one idea.
Computer Vision
The course that led to SPARK.
How it started
In eighth grade I was nearly kept out of the advanced Computer Science elective, the one that taught Java, because my mathematics marks were short, even though my Computer Science marks were good. I made my case for three days, to the head of the department and then in the principal's office, and they let me in. I thought computing would be my way out of mathematics. It turned out to be the opposite.
Papers and awards
Paper
Smart Parking System Using Computer Vision Based Parking Spot Detection and IoT Based Real Time Monitoring
International Conference on Computing and Communication Networks (ICCCNet-2025), hosted by Manchester Metropolitan University1–3 August 2025
The research paper from SPARK, published and peer-reviewed. I co-authored it and presented it remotely.
20241st place, Barclays Hack-O-Hire Quant Hawk. One of 27 teams selected from over 1,000.
2023Silver Medal, SRM Research Day Deshatan, the accessibility bulletin.
2022Top 10, Mozohack 3.0 Replete, at SRMKZILLA's hackathon.
2022Finalist, Smart India Hackathon internal selection Final round of SRMIST's internal selection for the national hackathon.
Certifications
AWS Cloud Foundations
AWS Machine Learning Foundations
AWS Data Engineering
College organizations and communities
Through college I belonged to eight student communities at SRM, always on the technical side.
Oct 2021
Alexa Developers SRM
Technical Associate, Oct 2021 – Jul 2022 Technical Executive, Jul 2022 – Jan 2023
Ran the technical side of our Alexa Skills and GraphQL events, including the logistics.
Wrote the backend for the club's event portal and kept it open source. It is Express and TypeScript over Supabase (Postgres), with AWS SES and S3, deployed on Heroku and Azure.
Looked after the backend teams for both events. Registration, RSVP emails through SES and QR-code attendance were all REST APIs.
Jan 2022
Microsoft Learn Student Ambassadors SRM
Technical Associate, Jan 2022 – Jul 2023
Contributed to a web app that sends files straight from one browser to another. It is TypeScript with peer.js, and the transfers are encrypted with crypto modules, so no server ever has to store the files.
May 2022
IEEE Student Branch, SRMIST
Web and App Developer, May 2022 – Nov 2022 Research and Development Associate, May 2022 – Jan 2023
Talked through how a regression-based machine learning model could fit into the way assignments are handed out across courses in a college, to make the process more efficient and more consistent between courses.
May 2022
Next Tech Lab, Tesla Lab
Technical Associate, May 2022 – Nov 2022 Technical Member, Nov 2022 – Oct 2023
Built the backend for the lab's product and team website with Express and TypeScript, using Supabase buckets for storage.
Jun 2022
SRM Alumni Affairs
Technical Team Member, Jun 2022 – Oct 2023
Sep 2022
TEDxSRMIST
Technical Associate, Sep 2022 – Sep 2023
Built the blog website for the TEDx event in TypeScript, with Next.js and Tailwind.
Oct 2022
SRMKZILLA
Technical Team Member, Oct 2022 – Oct 2023
Oct 2022
Apple Developers Group SRM
Technical Associate, Oct 2022 – Oct 2023
Skills
Languages
Python
Java
C
C++
JavaScript
TypeScript
ML and AI
PyTorch
Hugging Face
LangChain
LangGraph
NumPy
Pandas
Anthropic API
Backend
Spring Boot
Express
REST
GraphQL
Frontend
React
Next.js
Angular
HTML
CSS
Data
PostgreSQL
MySQL
MongoDB
Firebase
Supabase
Cloud and practice
AWS
Azure
CI/CD
DevOps
MLOps
Agile
Scrum
Outside work
Things that are not on a CV.
Piano
I hold Grade 4 Piano and Grade 3 Theory of Music, both with Distinction, from Trinity College London. I still play. Work makes the time harder to protect, and I still find it. It is how I stay level.
Training
I train and keep to my diet every week, and I do not renegotiate it. Nothing comes from one good day. It is months of ordinary days adding up, and the session that counts is the one after the day you skipped.
Debate and Model UN
I did debate and Model UN at Bishop Cotton, and kept up MUN at Amaatra. Both reward understanding a position well enough to state it fairly before taking it apart.
Courts
Basketball at both schools, and volleyball at Amaatra.
Chennai at eighteen
I left Bangalore at eighteen for four years in Chennai, and stayed on there for the training programme that opened my first job.
SPARK is the parking system I built in my final year at SRM. A camera finds the free spots, so a car park does not need a sensor in every bay. It grew out of my Computer Vision course, first as a semester project and then as my major project, and it became a paper at ICCCNet-2025.
I wanted this portfolio to feel like that project instead of only listing it. So the site is a small desktop, spark.os, sitting over a city. The car park in the middle is SPARK's: my projects are parked in its bays, and the camera view on the desktop draws boxes around cars and empty spaces the way a camera-based parking system would. The rest of my work lives in the buildings around it.
The idea of a portfolio you can drive around comes from Bruno Simon's site, and the desktop comes from Tommy Biglow's. I liked both and wanted them in one place.
The city, the car, the sounds and the icons are all generated in code. There are no image or audio files.