messy inputs to working tools

Ruby

Aadya Singh

Data science, software systems, and field notes with receipts.

I build practical data and ML tools, APIs, and interfaces that turn messy data problems into usable products.

View ProjectsGitHubOpen PRsProfilesResume

featured project

CabPilot ML

A full-stack taxi booking system that connects product flows, backend persistence, location parsing, and ML inference.

state changes recorded here

01Maps Input

Parses Google Maps links and turns messy pickup/dropoff text into usable trip coordinates.

02API + Auth

Separates rider and driver flows through FastAPI endpoints, JWT auth, and persisted trip state.

03Fare Model

Scores trips with a trained fare estimator and exposes the estimate transparently in the booking flow.

04Dispatch View

Keeps bookings, driver actions, and trip history inspectable instead of hiding the system in a notebook.

model signalFare estimate with error metrics

R² 0.742, MAE $2.70, RMSE $7.79.

edge casemessy maps links

problem solved

A taxi booking demo becomes a real software system, not just a fare-prediction notebook.

CabPilot connects rider booking, driver actions, authentication, trip persistence, Google Maps parsing, and a trained ML fare model through a typed FastAPI backend and Next.js interface.

11FastAPI endpointsR² .742fare model10parser tests
Next.jsFastAPIPostgreSQLSQLAlchemyML
View SystemGitHub

selected projects

Live from GitHub.

flagship build

CabPilot ML

A full-stack taxi booking MVP with rider/driver flows, JWT auth, trip persistence, Google Maps parsing, and transparent ML fare prediction.

R2 0.742 fare model + 11 API endpoints
Python0 starsUpdated Aug 8, 2026
Next.jsFastAPIPostgreSQLSQLAlchemyML
Open

production mlops

MLOps Drift Monitor

A predictive-maintenance system with model training, FastAPI machine-failure serving, PostgreSQL prediction logs, and sensor drift alerts.

Machine failure risk + sensor drift
Python0 starsUpdated Aug 8, 2026
PythonFastAPIScikit-LearnPostgreSQLDockerMLOps
Open

portfolio stopper

AgentOps Support Automator

A multi-agent task automation service for technical support triage, patch drafting, state history, and human-in-the-loop approval.

Stateful agent orchestration
Python0 starsUpdated Aug 8, 2026
PythonFastAPILangGraphPostgreSQLDockerNext.js
Open

retrieval ai

Medical RAG System

A healthcare-focused retrieval workflow for grounding medical-style questions in source documents and returning structured answers.

Grounded medical retrieval
Python0 starsUpdated Aug 8, 2026
PythonRAGLLMsVector Search
Open

data product

Australian Rental Insight Copilot

A data-driven assistant for exploring Australian rental market signals through analysis, retrieval, and an applied AI interface.

Local data insight system
Python0 starsUpdated Aug 8, 2026
PythonData AnalysisAIAustralia
Open

github repo

Aadyasingh55.github.io

A public GitHub project from my software, data, and AI portfolio.

Public build
TypeScript0 starsUpdated Aug 8, 2026
TypeScript
Open

github repo

Aadyasingh55

GitHub profile README for Aadya Singh: applied AI, MLOps, retrieval engineering, and software systems.

Public build
Code0 starsUpdated Aug 8, 2026
Codeapplied-aidata-sciencegithub-profilemlops
Open

github repo

Experiment Hub

Product experimentation analytics app with conversion lift, p-values, confidence intervals, SRM checks, power, and segment analysis.

Public build
Python0 starsUpdated Aug 8, 2026
Pythonab-testinganalyticsdata-scienceexperimentation
Open

github repo

Claimscope

Scientific claim verification prototype with document retrieval, evidence matching, stance classification, and evidence graph exploration.

Public build
Python0 starsUpdated Aug 8, 2026
Pythonclaim-verificationevidence-graphmachine-learningnlp
Open

Project archiveThis section is connected to GitHub as the source of truth.

Open ArchiveView PRs

what i build

Capabilities, grouped by outcome.

No wall of tools. Just the kinds of systems I can reason about and ship.

Intelligent Applications

ML and LLM features wired into real product flows.

Backend Systems

Typed APIs, auth, persistence, schemas, and deployment structure.

Machine Learning

Feature design, baselines, evaluation, and practical model behavior.

Data Engineering

Cleaning, shaping, documenting, and making data reusable.

Frontend Interfaces

Interfaces that make technical workflows understandable.

APIs & Databases

REST endpoints, SQLAlchemy models, and PostgreSQL-ready design.

experience

Education, certification, and public work.

University of Sydney

Master's in Computer Science, specialising in Data Science & AI and Software Engineering.

IIT Roorkee

Post Graduate Certificate in Data Science, Machine Learning, and AI.

GATE DA

Qualified in Data Science and AI with 92 percentile, AIR 4521, score 417.

Public technical work

Kaggle notebooks, GitHub projects, and Analytics Vidhya articles.

achievements

Signals that travel well.

Kaggle Datasets Expert

Rank 601/15,000+ with medals across datasets, notebooks, and discussions.

GATE-qualified

92 percentile in Data Science and AI, AIR 4521, score 417.

Technical writing

10+ Analytics Vidhya articles with multiple top-10 Blogathon rankings.

Open-source practice

Shared intelligent tools including medical NLP and summarization applications.

public profiles

Signals outside this site.

A few public trails across code, datasets, and writing.

code + open source

GitHub

31yearly contributions

Public repositories, portfolio builds, and external pull requests.

Repos and PRs refresh from GitHub.Open Profile

datasets + notebooks

Kaggle

Expertdatasets tier

Public data work, notebooks, discussions, medals, and dataset rankings.

Global datasets rank ~520.Open Profile

technical writing

Analytics Vidhya

11articles published

Applied AI, NLP, retrieval, summarization, and generative AI writing.

31 claps across published work.Open Profile

about

I like the space where data becomes usable.

I am currently pursuing a Master's in Computer Science at the University of Sydney, specialising in Data Science & AI and Software Engineering. My work sits between software engineering, applied analytics, and machine learning.

I am drawn to problems where the hard part is not only the model, but the surrounding system: the API, the database, the interface, and the evidence that the thing actually works.

Ruby

Let's build something useful from something messy.

LinkedInGitHubKaggleArticlesResume