Shubh
I build the quiet machinery behind reliable AI and honest dashboards — data pipelines, evaluation protocols, and reports that tell the truth.
From circuits to signals to sense.
I'm a data professional working at the meeting point of data engineering, analytics, and AI / LLM evaluation. My work is about making information trustworthy — moving it cleanly, questioning it rigorously, and presenting it so decisions get easier, not noisier.
I started in Electronics & Communication Engineering, which taught me to think in systems: inputs, transformations, noise, and the signal you're actually trying to protect. That mindset carries directly into pipelines and model evaluation today.
Electronics & Communication Engineering
Systems thinking, signal integrity, and an engineer's respect for what can go wrong between input and output.
Anudip certification & self-directed projects
Built practical data skills — SQL, spreadsheets, BI tooling — through structured certification and hands-on project work.
Ethara.ai — AI data operations
LLM output evaluation, prompt engineering, annotation, data ingestion pipelines, QA protocol design, and mentoring interns through the same standards.
Data analytics & AI evaluation
Focused on roles where data quality meets model quality — analytics with an evaluator's skepticism built in.
Things I've shipped and shaped.
LLM Output Evaluation & QA Protocols
Designed and ran evaluation workflows for large-language-model outputs — rubric design, annotation guidelines, inter-rater consistency, and QA protocols that kept quality measurable as the team scaled. Trained and managed interns to apply the same standards.
Data Ingestion Pipelines
Built and maintained ingestion pipelines feeding annotation and evaluation workflows — validation at the door, consistent schemas, and fewer surprises downstream.
Power BI Dashboards & Excel Models
Interactive reports and spreadsheet models built for clarity first — measures that answer real questions, visuals that don't decorate, and layouts a stakeholder can read in thirty seconds.
This Website
A hand-coded canvas animation — ninety silk strands sharing one flow field, a twinkling starfield with diffraction spikes, and nebula glows — wrapped in the editor you're reading now. No frameworks, no libraries.
A toolkit for trustworthy data.
Data Engineering
- Data ingestion pipelines
- SQL & data modelling
- Validation & data quality checks
- Schema & workflow design
Analytics & BI
- Power BI — reports & DAX
- Advanced Excel modelling
- KPI design & storytelling
- Exploratory analysis
AI / LLM Evaluation
- LLM output evaluation
- Prompt engineering
- Annotation & rubric design
- QA protocol development
Ways of Working
- Intern training & management
- Process documentation
- Systems thinking (ECE roots)
- Detail-first troubleshooting
Open a channel.
Open to Data Analyst and AI / LLM Evaluation roles, freelance analytics work, and interesting conversations about data quality.
Kraven streams — channel analytics.
Kraven streams
The ultimate gaming zone — streams and uploads on Valorant, GTA V, BGMI, and Free Fire, with a community on Discord.
What the channel covers
Live streams, gameplay highlights, and regular uploads across four titles — tactical FPS, open-world, and battle royale.