I build systems that make data useful.

I'm Stephen Baraik, a data scientist and engineer building reliable infrastructure, machine learning systems, and AI products that hold up beyond the notebook.

Explore the work Work with me
Navi Mumbai, IndiaRead my resume
Stephen Baraik
Stephen BaraikData scientist & engineer
Building with purpose.
From data to a working product.

Selected work

Real problems. Working systems.
A look at what I've built end to end.

Honest model evaluation
Before the leakage fix99.7%
After the leakage fix88.3%

A lower number.
A model you can trust.

Why lower is better

Identified and fixed label leakage across 3 models, correcting an inflated 99.7% rainfall accuracy down to a genuine, trustworthy 88.3%.

Stormwatch AI / rainfall prediction

Stormwatch AI

Extreme Weather Prediction System

Trained ML models forecasting cyclones (98%), heatwaves (99%), and rainfall (88%) across 15 Indian cities on 90,000+ real records.

PythonXGBoostFastAPIMLflowDockerPySparkGitHub Actions
Engineering notes
  • Identified and fixed label leakage across 3 models, correcting an inflated 99.7% rainfall accuracy down to a genuine, trustworthy 88.3%.
  • Deployed a 5-endpoint FastAPI service with an MLflow registry and Docker Compose, processing 90k+ records through a PySpark ETL pipeline.
  • Automated CI/CD with GitHub Actions, backed by 80 passing pytest tests and KS-test drift monitoring.
Read the case study GitHub
PaisaPro

From market data
to an informed decision.

Market dataForecast & riskAI advisor
502 NSE/BSE stocks / 35+ API endpoints

PaisaPro

AI Investment Advisory Platform

An AI-powered forecasting and risk engine covering 502 NSE/BSE stocks, combining classical quant models with an LLM-driven advisor.

PythonFastAPIReactTypeScriptSupabaseGroq/Llama 3.3Docker
Engineering notes
  • Built forecasting and risk models using ARIMA, GARCH(1,1), and the Fama-French 4-factor model.
  • Trained a walk-forward-validated ML ensemble on 20+ OHLCV features for 30-day price prediction.
  • Built a FastAPI backend with 35+ endpoints across 19 services, powering an agentic Llama 3.3 advisor with live tool calls over SSE.
  • Shipped an 18-page React SPA on Netlify, backed by Supabase and a Hugging Face Spaces Docker deployment.
GitHub Live demo

ChartCraft

Open-Source Python Dashboard Framework

An open-source Python dashboard library built to rival Power BI and Tableau, with 18+ chart types and 11 themes.

PythonApache EChartsSSE
Engineering notes
  • Engineered the core on Python's standard library and Apache ECharts 5.5, with zero external dependencies.
  • Built a drag-and-drop canvas synced with the underlying Python in real time via bidirectional, AST-based codegen.
  • Shipped SSE-based live streaming with independent per-chart refresh, plus export to HTML, Jupyter, Docker, and PDF.
GitHub

Vivah Verse

Wedding Marketplace Platform

A two-sided wedding marketplace connecting couples and vendors, with role-based workflows for couples, vendors, and admins.

Next.jsTypeScriptSupabaseTailwind CSS
GitHub Live demo

Bitcoin Data Pipeline

A fully containerized real-time and historical data pipeline for Bitcoin market data.

PythonApache AirflowPostgreSQLDocker
Engineering notes
  • Automated a Python module pulling historical and real-time OHLC candles from the Binance API.
  • Built two Airflow DAGs: one for historical backfill, one for real-time ingestion running every minute.
  • Persisted candle data in PostgreSQL with idempotent inserts to prevent duplicate records.
  • Containerized the full stack (Airflow, PostgreSQL, pgAdmin) with Docker Compose for reproducible local setup.
GitHub

RL Bitcoin Trader

Reinforcement learning experiments applying algorithmic trading strategies to Bitcoin markets.

PythonStable-Baselines3Reinforcement Learning
GitHub

Zoc Chatbot

Experimental RAG Lead Qualification Bot

An experimental retrieval-augmented generation chatbot built for automated lead qualification at Zoc Learnings.

FastAPILangChainChromaDBNext.js
Live demo

Let's build something useful.

Practical AI and websites for businesses.
Direct communication. Work that ships.

Elnova AI

Founder

Elnova AI is an AI consulting practice I run alongside my full-time work, helping businesses figure out where AI actually fits into their operations and then building it for them — from automation and internal tooling to custom LLM-powered applications.

  • Advise SMBs and founders on practical, high-ROI AI adoption: automation, chatbots, internal copilots, and data workflows, rather than AI for its own sake.
  • Design and build custom AI solutions end-to-end: scoping, model/tool selection, prototyping, and production deployment.
  • Bring the same production-grade engineering standards from my data science work — clean pipelines, monitored systems, and honest evaluation — into client projects.
AI Strategy ConsultingChatbot & RAG DevelopmentWorkflow AutomationLLM Integration
Talk about a project

Freelance Web Development

SMB Websites

Alongside Elnova AI, I take on freelance projects building websites for small and medium businesses that need a clean, functional web presence without enterprise-agency overhead.

  • Design and build fast, modern websites for SMBs — landing pages, business sites, and small e-commerce — end to end.
  • Handle the full stack: design, frontend build, deployment, and basic SEO/schema setup.
  • Focus on quick turnaround and direct communication, since most clients are small teams who need things shipped, not managed.
React/Next.jsTailwind CSSSupabase/FirebaseNetlify/Vercel
Talk about a project

Beyond the notebook.

A practical approach to ambitious systems.

Read my resume

I'm a Data Management Analyst at Zoc Learnings, where I work at the intersection of data engineering and applied AI — building CRM data pipelines, automating reporting, and prototyping RAG-based tools for the business. I graduated from NMIMS University with a B.Sc. in Data Science in 2026, and I hold a DataCamp Associate Data Engineer certification.

Outside my day job, I build end-to-end products: forecasting engines, dashboard frameworks, real-time data pipelines, and full-stack applications. I care about the unglamorous parts of data work — clean pipelines, honest model evaluation, and systems that don't fall over.

Alongside my day job, I run Elnova AI, an AI consulting practice helping businesses adopt AI in practical, high-ROI ways, and I take on freelance projects building websites for small and medium businesses.

I'm currently exploring opportunities based in Singapore, and I'm always open to conversations about data engineering, machine learning, and building useful products.

Experience

Jan 2024 – Present

Remote

Zoc Learnings

Data Management Analyst

  • Centralized data from 10 sales staff into per-person CRM sheets, pipelined into a single compliant master Zoc CRM.
  • Automated weekly payment-pending and monthly sales reports via Apps Script, auto-generating HTML emails for stakeholders.
  • Prototyped an experimental RAG chatbot (FastAPI, LangChain, ChromaDB, Next.js) for automated lead qualification.
  • Audited company websites using data science techniques, fixing 115 broken pages and adding sitewide schema coverage.

The toolkit

From the first query
to the production pipeline.

Programming

  • Python
  • SQL
  • R
  • Bash

Machine Learning & AI

  • Scikit-learn
  • TensorFlow
  • PyTorch
  • MLflow
  • LangChain (RAG)
  • Stable-Baselines3
  • LLM Integration

Data Engineering

  • Apache Spark
  • Hadoop (HDFS, MapReduce)
  • Airflow
  • ETL Pipelines
  • Docker

Backend & Web

  • FastAPI
  • React
  • Next.js
  • Supabase
  • Tailwind CSS

Databases & Cloud Data

  • PostgreSQL
  • MongoDB
  • Supabase

Analytics & Visualization

  • Pandas
  • NumPy
  • Matplotlib
  • Seaborn
  • Tableau
  • Power BI

Developer Tools

  • Git
  • Jupyter
  • WSL
  • LaTeX
  • Windows
  • Linux

The foundations

2023 – 2026

NMIMS University

B.Sc. in Data Science

Relevant coursework: Machine Learning, Data Warehousing, Statistical Methods, Big Data Analytics, Python Programming.

2025

DataCamp

Associate Data Engineer Certification

Good work starts
with a conversation.

Building a team, an AI product, or your next website? I'd like to hear about it.

I'm also exploring data science and engineering opportunities, with a particular interest in Singapore.

stephenbaraik@gmail.com
LinkedIn GitHub