B.Tech — Computer Science & Engineering
GLS University
System Design · Machine Learning
The journey · 3 chapters
Experience.
A chronological record of the work — from machine learning internships to co-founding an AI startup while studying at GLS University and IIT Madras. The constants: a bias for shipping, and a stubborn love of the details.
-
Co-Founder
Colrnx
Building Colrnx, an AI-verified talent infrastructure platform that transforms developers’ real production workflows into authentic hiring signals through continuous telemetry and AI-driven evaluation.
- Secured early institutional support from Function1 Founders Hub (Dubai)
- Owning the platform end to end — telemetry pipeline, AI evaluation, and the product loop that turns real work into hiring signal
-
AI Engineer and Product Manager
Clan
Designed a Python-based AI content automation pipeline for an education platform, cutting content creation time from 6–7 hours to 2 hours per batch.
- Reduced content creation time from 6–7 hours to 2 hours per batch, saving 20+ hours weekly for the content team
- Optimized backend workflows and databases managing 750+ video content pieces, improving delivery reliability
- Managed database systems serving 100+ education creators across the platform
-
ML Researcher
Prodigy Infotech
Built a Python-based ML pipeline for preprocessing, feature engineering, and statistical validation across 3+ baseline models.
- Applied KNN, Random Forest, and Logistic Regression for feature selection and importance analysis on structured datasets
- Incorporated multiple imputation techniques to strengthen prediction reliability across baseline models
Education
B.S. — Data Science
IIT Madras
Data Science · Artificial Intelligence
Research
Data Preprocessing Pipelines in Machine Learning
Authored a review paper on AI data preprocessing across machine learning, large language models, and agentic AI systems — synthesizing insights from 50+ research works on scalable and multimodal pipeline strategies.
Skills
Languages
- Python
- JavaScript
- SQL
- C++
Frameworks & Libraries
- React
- Next.js
- Vue
- Node.js
- FastAPI
- Flask
- Tailwind CSS
AI & ML
- PyTorch
- Transformers
- Hugging Face
- RAG
- ChromaDB
- FAISS
- OpenCV
- Librosa
Databases
- PostgreSQL
- Supabase
- Redis
- SQLite
- Firebase
- MongoDB
Tools & Platforms
- GitHub
- Git
- Docker
- Jira
- Linux
- GitHub Actions
How I work
Write it down
Decisions without a written reason are opinions with a timestamp. I keep the notes so the team can argue about the right things.
Ship in six weeks
If I can't see a version of the thing that ships in six weeks, I don't understand the problem yet. Small and honest beats large and imaginary.
Let the data speak
Models fail silently; products fail loudly. I evaluate against real workflows, not benchmarks — and I'd rather a system that is honest about its uncertainty than one that's confident and wrong.