Hey, I'm
I build intelligent systems that feel trusted
I turn messy data, raw language, and unclear workflows into NLP, RAG, and Agentic AI tools – useful systems people can understand, trust, and actually use.
It started with a taxi app, not a textbook.Everything since – the CS degree, the M.Tech, the tools – came out of one question about how systems decide.
Don't just aspire to make a living – aspire to make a difference. – Denzel Washington
Hi, I'm Aradhya Pavan – I build useful intelligent systems people can actually understand, trust, and use.
A visit to Bengaluru – watching Uber/Ola match riders, routes, demand, and timing in real time – pulled me into B.E. Computer Science. Later, my M.Tech in AI & ML from BITS Pilani pushed that curiosity toward NLP, RAG, and agentic AI systems that move from POC/MVP to responsive, practical web apps.
On the side: small tools that cut repetitive work, online hackathons, and a focus on design, use-case fit, UX, attention to detail & cost. Big Formula 1 fan – the live strategy, telemetry, and split-second decisions genuinely feed how I think about systems.
Used, not listed.Every tool here is one I have shipped something with – not one I read about.
I build language systems that can retrieve the right context, reason through a workflow, and return something a real user can trust – from RAG and agentic tools to fine-tuned models shipped as clean, usable web experiences.
Python
MLflow
PyTorch
Keras
Scikit-Learn
NumPy
Pandas
SpaCy
NLTK
Matplotlib
Seaborn
LangChain
Hugging Face
PhiData
FAISS
AWS
Azure
HTML
CSS
JavaScript
Flask
PHP
REST APIs
Swagger
PostgreSQL
MySQL
Git / GitHub
Docker
Heroku
Streamlit
One flagship, two I still think about.Each has a decision in it I got wrong the first time. The case studies say which.
Agentic AI
Ask a database questions in plain English and get correct, runnable SQL back – the agent reads your schema, picks the right dialect, and validates every query before it runs.
NLP · Education
Learn NLP by doing – from text preprocessing to transformers and semantic search, with every concept paired with runnable, visualised examples in the browser.
M.Tech
Dissertation
Turns large volumes of open-text feedback into structured, actionable insight – a multi-agent pipeline extracts entities, resolves references, and streams downloadable results.
Quieter by design.The full archive sits after the flagship work so recruiters can scan without losing the main story.
Verified, not decorative.Microsoft, AWS, DataCamp and GitHub – each one links to its issuer, and they sit deliberately after the work.
The reps nobody sees.Open source, experiments, structured learning and competitive problem solving – the practice behind the shipped work.
CodeChef
HackerRank
LeetCode
Codeforces
Coursera
DataCamp
Codecademy
GfGNo form. Just email.I'm open to AI/ML engineering roles, freelance builds, and collaborations where practical AI, clear UX, and real deployment matter.
