I build machine learning that actually ships. The maths and the models underneath, the GenAI work on top, and I teach both.
I am a final year Data Science student at IIT Madras, and I work as a Data Scientist and SME at Scaler. I build, test and ship machine learning, deep learning and LLM work. Under all of it sits a lot of statistics, optimisation and plain problem solving. The other half of my job is turning hard technical ideas into something hundreds of learners can follow.
About
소개I like the whole of machine learning, not just one end of it. The maths and the stats on one side. Getting a model to actually work for real people on the other.
At Scaler I write the AI Engineering coursework. That means taking hard topics like LLMs, RAG and agents and turning them into lessons people can actually follow. I also build RAG pipelines and LLM agents that take the boring repeated work off our own team. At Uncia (IITM Research Park) I built models for anomaly detection, association rule mining and time series forecasting, and then helped get them running live.
Most of what I build runs end to end. That means the data checks first, then the training and the evaluation, and then the deployment and the monitoring that tells me whether the thing still works next week. I care about how a problem gets solved and not only about whether the final number looked good, which is the part of this work I enjoy most.
Three years of teaching at IIT Madras changed how I work. If I cannot explain a choice in plain words, I usually do not understand it well enough yet. So I keep my code easy to read, I write down why I picked one thing over another, and I only report numbers I could defend to someone who did not build the thing.
- /Maths first. Statistics, optimisation, algorithms and the problem solving under the models
- /Deep learning with PyTorch, TensorFlow, transformers, fine tuning and LoRA
- /Classical ML with scikit-learn, XGBoost, LightGBM, forecasting and anomaly detection
- /GenAI work like RAG pipelines, LLM agents, prompting and evaluation
- /Getting it shipped with FastAPI, PostgreSQL and pgvector, Docker and CI/CD
Experience
경력Selected projects
작업Technical stack
도구Education
학력Beyond the work
취미Away from work, most of my free time goes to Korean film, music and food. That is where the typography and the quiet look of this site came from.
The same habit shows up in how I learn. I take a thing apart and build the interesting bit again myself. That is how ANN from Scratch started. I wrote the activation functions, the weight setup, the loss functions and gradient descent in plain NumPy instead of trusting the library to do it for me. I read papers the same way.
Get in touch
연락Open to AI and ML engineering roles from September 2026.