Agentic AI for engineering workflows
Designed and built an agentic AI system that automates repetitive engineering workflows, from prototype to production.
Software Engineer @ Couchbase · Bangalore, India
Rahul
Vijayakumar
> backend · agentic AI · infrastructure
I build systems that actually get used in the real world, especially where backend engineering meets AI.
scroll ↓01about
I'm a software engineer at Couchbase, where I build and maintain an agentic AI system that runs in production and takes repetitive work off other engineers' plates. I also debug and fix complex distributed systems, both on-prem and across the major cloud providers.
I got deep into LLMs at university, fine-tuning models and deploying them for real users. Self-hosting that work is what pulled me into DevOps and infrastructure: networking, containers, virtualization and storage. Before Couchbase, I spent a year as a backend and DevOps project intern at the Ministry of Electronics and Information Technology, where I built the backend for an AI learning portal and ran a GPU server for fine-tuning and serving open-source models. Researched and implemented fine-tuning techniques for small language models to create domain-specific experts. I also mentored the next generation of interns.
At VIT Chennai I was Technical Co-Head of the Linux Club, where I led the team that built the club's event platforms, and I co-authored two papers on deep learning for medical imaging, published by IEEE and Springer Nature.
- languages
- Python
- Go
- JavaScript
- C++
- Bash
- backend & data
- FastAPI
- Django
- Flask
- LangGraph
- PostgreSQL
- MySQL
- MongoDB
- Couchbase
- devops & cloud
- Docker
- Nginx
- Linux
- GitHub Actions
- AWS
- Azure
- GCP
- ml
- vLLM
- TensorFlow
- OpenCV
02selected work
at Couchbase
at Ministry of Electronics and Information Technology
AI Learning Portal
Developed a scalable Python backend for an AI-powered learning portal that enhances student engagement through interactive, personalized content. Conducted research on domain-specific language model fine-tuning, creating specialized LLMs for targeted educational applications.
- Python
- AWS
- Amazon RDS
GPU server for AI training and inference
Set up and managed a GPU server for AI training and inference, used to fine-tune open-source models and serve them for inference.
- CUDA
- Fine-tuning
- Inference
publications
Alzheimer's Disease Classification Using CUDA-Accelerated DenseNet201 and VGG16
IEEE · 2025A GPU-accelerated, multi-branch network that combines modified DenseNet201 and VGG16 to classify the stage of Alzheimer's disease from brain MRI scans. CUDA optimization across data loading, augmentation and training cuts training time, and the model reaches 98% accuracy across four stages.
- CUDA
- DenseNet201
- VGG16
- MRI
Augmented Transfer Learning for Skin Cancer Detection: Enhancing Accuracy Using Edge Detection
Springer Nature · 2025Transfer learning with CNNs, enhanced with edge detection, to classify skin lesions as malignant or benign so skin cancer can be caught earlier.
- Transfer learning
- CNNs
- Edge detection
at university
LLM chatbot for the student portal
20k+ studentsFine-tuned LLMs and served them with vLLM to power a chatbot built into the university's student portal. Used by 20,000+ students and ran reliably, which taught me a lot about building for scale.
- Python
- vLLM
- LLM fine-tuning
Self-hosted infrastructure
We self-hosted most of the chatbot's stack, which pushed me into DevOps: networking (port forwarding, NAT, gateways), containers, virtualization and RAID storage. Before that, I turned an old PC into the server behind the Linux Club forums, used by 200+ members.
- Linux
- Docker
- Proxmox
- Nginx
Linux Club event platforms
Technical Co-HeadLed the team that built the club's platforms for event registrations, attendance tracking, recruitment and CTF competitions. We deployed them on AWS and Azure and kept them up under real event-day traffic.
- AWS
- Azure
side projects
Recruitment Platform
A recruitment platform for applicant shortlisting, interview score tracking, and feedback management. Built for Linux Club VIT Chennai's 2023 recruitment drive, it streamlined the hiring process for hundreds of applicants across multiple departments.
- Python
- SQL
MovieTeleparty
A watch-party web app: one host controls playback, friends join with a session code, and everyone's video position stays in sync every second.
- Flask
- JavaScript
- Docker
GroupFinder
Helps VIT Chennai students discover and join the right WhatsApp groups.
- Python
- Vercel
QR code attendance
A Flask app that generates and scans QR codes to take attendance and keep the records.
- Python
- Flask
TechTrivia answer portal
Answer submission portal for Zero Bugs Club's TechTrivia event, with a separate admin portal for reviewing answers.
- Python
- Vercel
03experience
- 2025
Jul 2025 – present
Software Engineer · Couchbase
Designed, built and now maintain an agentic AI system that automates repetitive engineering workflows in production. Alongside that, debugging and fixing complex distributed systems, on-prem and across the major cloud providers.
Jan 2025 – Jun 2025
Software Engineer Intern · Couchbase
- 2023
Dec 2023 – Dec 2024
Backend Developer and DevOps Project Intern · Ministry of Electronics and Information Technology
Remote internship. Built the backend for an AI learning portal, and set up and managed a GPU server for fine-tuning and serving open-source models. Worked with AWS, including Amazon RDS, and mentored the next generation of interns.
Nov 2023 – Dec 2023
Software Developer Intern · Chakaralaya Analytics Pvt Ltd
Software development internship in Chennai, working with databases and Docker.
May 2023 – Jan 2024
Technical Co-Head · Linux Club, VIT Chennai
Led development of the club's registration, attendance, recruitment and CTF platforms, and ran the self-hosted server behind the club forums.
Apr 2023 – Dec 2023
Web Developer · Microsoft Innovations Club, VIT Chennai
- 2022
Oct 2022 – May 2023
Web Developer · Hack Club, VIT Chennai
Worked on the club's main website and event tools, like the poll site for the Yellow Filter event.
Jun 2022 – May 2023
Technical Advisory Team · Linux Club, VIT Chennai
Member of the club's technical department.
Feb 2022 – May 2023
Member · Zero Bugs Club, VIT Chennai
Built the answer portal for the club's TechTrivia event.
04education
- 2021
2021 – 2025
B.Tech, Computer Science and Engineering · VIT Chennai
CGPA 9.00.
2021
Class 12, CBSE · Bishop Moore Vidyapith
All India Senior School Certificate Examination (AISSCE): 97.6%
- 2019
2019
Class 10, CBSE · Bishop Moore Vidyapith
All India Secondary School Examination (AISSE): 96.8%