Press 1 to 7 to jump to any page.
// projects

Selected work

A selection of full-stack, AI, ML, and data projects. Each began with a practical problem and ended with something testable, useful, and ready to improve.

RAGComputer visionInternship

Applied AI at QuantAI

Remote Data Science internship for a company based in Auckland, NZ. Built RAG pipelines, a YOLOv8-based video analytics system, and a sales analytics dashboard for a live client engagement.

RAG pipelinesYOLOv8 analyticsBI dashboards
ReactSupabaseFull-stack

CivicSpot

A civic issue-reporting platform that lets citizens report and track city problems with AI-verified photo evidence, an interactive map, and an admin dashboard.

Photo verificationInteractive mapAdmin dashboard
FastAPIGroqDeploying

Querify

An AI-powered BI platform that turns CSV/XLSX uploads into dashboards, forecasts, and natural-language queries. No SQL required.

NL queryingAnomaly detectionForecasting
LLMsFAISSLocal / offline

Local Llama 3 8B Q&A System

Offline RAG-powered Q&A and summarisation built with a quantized Llama 3 8B model, FAISS vector retrieval, and local inference - evaluated with structured answer validation and a 25+ test pipeline suite.

FAISS retrieval4-bit NF4 quantized~38 tok/s on RTX 3090
FirecrawlSupabaseNot deployed

Sentiment Spark

An AI sentiment-analysis tool for pasted text or any web-page URL, scraping YouTube, Reddit, blogs, and news via Firecrawl.

Sentence breakdownKeyword extractionSummary
YOLOv8OpenCVFull-stack

Video Detection

A full-stack video analytics app that detects objects in uploaded videos and infers the activity taking place, tuned through frame sampling and confidence thresholding.

Object detectionActivity inferenceReact + Material UI

Contact

Have an AI problem worth solving?

For roles, projects, and thoughtful conversations.

© 2026 Shreesh Dwivedidesigned & built with curiosity.

Contact

My local time: loading…

Email

Always happy to help.

Compose

Stay in touch

I'm most responsive on LinkedIn.