working notebook
DATAIDEAS
I build data products end-to-end — from ETL pipelines and dashboards to RAG systems and agentic LLM workflows.
Data Scientistfield notes
Open to Data Science & AI roles
Hello, I'm Bhadresh V K
Data Scientist
I build data products end-to-end — from ETL pipelines and dashboards to RAG systems and agentic LLM workflows.
PythonMachine learningData products
- 4
- GitLab projects, one coherent platform
- 3
- data source types in a single ETL
- 50+
- automated tests across repos
- 2
- answer backends in the RAG system
About
Data scientist with an end-to-end mindset
Data science master's student (Web & Data Science, University of Koblenz) with hands-on experience in Python, SQL, and machine learning — plus applied work in GenAI: RAG architectures, embeddings, vector databases, and agentic LLM workflows.
I build data products end-to-end: automated collection, validated ETL pipelines, star-schema analytics, dashboards, and applied ML systems. My GitLab portfolio is one coherent market-intelligence platform — every repo tested, documented, and CI-ready.
margin note
Curiosity in. Clean data out.
The interesting part is often the handoff: the question, the data, the model, and the person who needs the answer.
Read the working notes
Programming
Python (Pandas, NumPy)SQLJavaScript / TypeScriptRShell
ML & Deep Learning
Scikit-learnTensorFlow / KerasPyTorchLSTM / GRUClassificationAnomaly DetectionTime Series
GenAI / LLM
RAG ArchitecturesPrompt EngineeringEmbeddingsVector DatabasesAgentic WorkflowsLangChainLLM / AI APIs
Data & BI
EDA & StatisticsHypothesis TestingData PipelinesMySQL / PostgreSQLPower BIMatplotlib / Seaborn / Plotly
Tools & MLOps
Git / GitHub / GitLabDockerCI/CDLinuxpytestAWS / Azure / GCP
Background
Experience & education
Experience
ML Intern (Python)
2023Zebo.AI × Verzeo
Predictive ML models, data pipelines, and automation in Python.
Certificate — ADAS & ML
2022Reynlab
Driver-assistance systems, sensor data analysis, model-based decision making.
Education
M.Sc. Web and Data Science
2025 — 2027 (expected)University of Koblenz
Focus: Data Science, Machine Learning, Statistics, Big Data, NLP.
B.Tech Computer Science & Engineering
2019 — 2023APJ Abdul Kalam Technological University
Focus: Algorithms, Data Structures, Databases, AI fundamentals.
Projects
End-to-end data platform
AI RAG Prototype
Local-first Retrieval-Augmented Generation document Q&A: embeddings, ChromaDB vector search, cross-encoder re-ranking, and cited answers via an offline extractive reader or any OpenAI-compatible LLM.
- Fully offline, no API keys required
- 50+ pytest tests + SQuAD retrieval eval
- CLI, Streamlit, and Python API interfaces
PythonRAGChromaDBSentence-TransformersStreamlitpytest
ETL Data Pipeline — EV Market Intelligence
Multi-source ETL pipeline (REST API + CSV + SQL) that extracts, validates, and transforms data into a star-schema analytical dataset with quality gates, exported as Parquet/CSV for Power BI.
- 3 source types, one clean model
- Star-schema output with upsert load
- Automated quality gates in CI
PythonETLSQLSQLAlchemyPandasParquet
Power BI Dashboard — Market Intelligence
EV market-intelligence dashboard on a clean star-schema model. Every KPI is verified in Python with unit tests before it becomes DAX; ships a Streamlit demo and a Power BI Desktop guide.
- KPI model tested in Python before DAX
- Star-schema fact + dimensions
- Interactive Streamlit preview
Power BIDAXStreamlitPython
Automation Bot — Market Intelligence Collector
Scheduled RPA-style bot that collects signals from RSS feeds, deduplicates them, and produces daily Excel reports — the ingestion layer that feeds the ETL pipeline and the dashboard.
- Headless scheduled runs (cron / launchd)
- Deduplication across sources
- Formatted Excel report output
PythonAutomationRSSExcelcron / launchd
A closer look
Selected work
Earlier experiments and coursework that still shape how I work with models and data.
2025
Time Series Forecasting & Anomaly Detection
EDA and modeling on weather/sensor data; trained and evaluated recurrent networks (LSTM/GRU) for forecasting with anomaly detection, performance metrics, and clear visualizations.
PythonTensorFlow / KerasPandasMatplotlibPlotly
2022
Music Recommendation via Facial Expression Recognition
Recommendation system that suggests music based on the user's facial expressions — image processing, classification, and API integration.
PythonOpenCVTensorFlowSpotify API
Field note17.12.20223 min read
How AI (GAN) can generate images?
How GAN generates Images
By GPTChat, it deserves the crown
GANImage GenerationDeep Learning
Read the note 
Field note17.12.20223 min read
Neural Networks
A neural network is a type of machine learning algorithm modeled after the structure and function of the human brain. It consists of a large number of interconnected processing units, called neurons, which work together to process and interpret input data.
By GPTChat
Neural NetworksMachine LearningDeep Learning
Read the note 
Field note17.12.20227 min read
Stable Diffusion
My questions ralated Stable Diffusion answered by ChatGPT
By Bhadru and GPTChat , Mostly ChatGPT
Stable DiffusionTransformersText-to-ImageDiffusion Models
Read the note Contact
Let's build something with data
Available for data science, ML, and analytics engineering roles. I'm always happy to talk about data pipelines, RAG systems, or dashboards — feel free to reach out.
Send an emailBased in Koblenz, Germany