Work

Proof that ZettaMetrics can build real systems.

The portfolio leads with full-stack product proof, then shows the broader archive across AI, ML, recommendation, and data science work.

Agentic RAG product

ShaileshGPT

A full-stack AI portfolio system with grounded chat, recruiter JD-fit analysis, lead capture, analytics, notifications, and PDF reports.

Shows the ability to turn a static profile into an interactive AI product with real backend workflows.

ReactFastAPIOpenAIRAGSupabaseSendGrid

Client EdTech platform

A full-stack math learning platform with admin, teacher, and student flows, DPS assignments, server-side scoring, timers, and result review.

Demonstrates product engineering across curriculum tooling, protected assessment flow, and operational dashboards.

Next.jsFastAPISQLTanStack QueryRole-based UX

Multi-agent research system

An orchestrated research assistant that clarifies queries, plans evidence gathering, searches, synthesizes, and can deliver reports by email.

Applied GenAI

A LangChain and OpenAI-powered question-answering project focused on practical LLM integration patterns.

Time-series forecasting

LSTM-based forecasting project using historical stock data, trend visualization, and predicted-versus-actual comparison.

Recommendation system

Content-based recommender using metadata, vectorization, and similarity scoring to surface relevant movie suggestions.

NLP recommendation

Course recommendation project using TF-IDF and cosine similarity over course descriptions and metadata.

Applied classification

Binary classification project predicting placement likelihood from academic and profile signals.