I'm a CS student (Data Science specialization) at Alva's Institute of Engineering and Technology, Mangalore. I work across the full stack — from training CNN models and building ETL pipelines to shipping web apps in Next.js and Go.
This past year I interned at Edunet Foundation (in collaboration with Shell India & AICTE), where I built an automated image classification system using transfer learning on a dataset of 5,500+ raw images and got it to 94% inference accuracy. I also won the NASA Space Apps Challenge (Galactic Problem-Solver, 2025) and our department hackathon in 2026.
Most of what I build lives somewhere between ML and systems — real-time RFID anomaly detection, NLP sentiment pipelines, local CI/CD simulation engines, or AQI forecasting models pulling from NASA APIs. I like problems where the data is messy and the latency constraints are real.
What I'm working on right now:
- YamlAnchor — a local-first CI/CD self-healing tool that scans your repo, generates GitHub Actions YAML, and validates it inside Docker containers before you ever push
- GrokCity — a real-time NLP sentiment analysis engine built on the Groq API, with a 3D visualization layer, keeping p99 latency under 50ms
Stack I actually use:
Projects worth looking at:
| YamlAnchor | CI/CD self-healing pipeline generator. Scans your repo, writes the YAML, runs it locally in Docker. Next.js + Go. |
| Garbage Classification AI | EfficientNetB0 classifier for 6 waste categories, 89.24% test accuracy. Gradio web interface included. |
| AeroBuy | Flask + MongoDB e-commerce store with 200+ products. Aggregation pipeline optimization cut query latency by 35%. |
| arthasetu-website | Financial advisory portal. Includes a precise SIP calculator, portfolio tracker, and wealth strategy dashboard. |
GitHub activity:
Get in touch: rakeshgr223@gmail.com · linkedin.com/in/rakeshgr18 · rakeshgr.netlify.app · +91 9035402276