Prince Maurya
Ai engineer /LLM engineer
New to Contra
Prince is building their profile!
RE:WILD — Choosing Imperfection
For the Envato Challenge, I created RE:WILD, an eco-luxury skincare identity built around the Human by Design theme.
The first direction looked polished, but too perfect. My Human by Design moment was rejecting that clean, generic result and choosing imperfection: warmer tones, uneven botanicals, tactile grain, restrained typography, and a calmer brand story.
AI created the options. Human taste created the meaning.
Canva design: https://www.canva.com/d/bxK34YUB9illchz
Note: My Envato Plus coupon showed as expired during production, so this submission includes the campaign concept, creative workflow, final design system, and prompt direction prepared for Envato AI/assets once access is restored.
@envato #envatochallenge #HumanByDesign #BrandIdentity #CreativeDirection #AIDesign
1
3
119
Title: DevMind - Autonomous AI Coding Agent
Description:
Built DevMind, a Claude Code-inspired autonomous coding agent that can plan and execute multi-step coding tasks inside a terminal.
I built it using Python, LangGraph, Anthropic API, and LangChain. The agent supports real-time token streaming and includes 6 security-hardened tools: file read, file write, file edit, bash execution, grep, and directory listing.
The hardest part was making tool execution safe and reliable. I added path-traversal protection, dangerous-command blocking, API-key masking, retry logic, token-bucket rate limiting, autosaved sessions, and latency metrics.
Outcome:
Validated with 156 tests across unit, integration, and security scenarios.
Tech:
Python, LangGraph, Anthropic API, LangChain, CLI Agents, Security, Testing
GitHub:
https://github.com/PRINCE2-AI/devmind
0
14
Title: Local RAG System - Private Document Q&A
Description:
Built a privacy-first document Q&A system that lets users upload PDFs or text files and ask questions over their own documents.
The system ingests documents, splits them into chunks, creates local embeddings with Ollama, stores them in ChromaDB, retrieves relevant context, and generates answers using a local LLM.
The hardest part was connecting the full RAG pipeline cleanly: loading, chunking, embedding, retrieval, and answer generation while keeping everything local.
Outcome:
A working private RAG app where no document data leaves the machine.
Tech:
Python, LangChain, ChromaDB, Ollama, Streamlit, RAG, Vector Databases
GitHub:
https://github.com/PRINCE2-AI/local-rag-ollama-chromadb
0
18
Title: OpenAI Autodata - Agentic Synthetic Data & Evaluation Pipeline
Description:
Built an agentic pipeline for generating and evaluating synthetic data using multiple LLM roles.
The system uses challenger, weak solver, strong solver, and judge agents to create difficult examples, compare model behavior, and accept only examples that pass score-gap and validation checks.
The hardest part was making the evaluation deterministic enough to trust. I added fail-closed validation, auditable JSONL outputs, cost controls, and test coverage so bad or incomplete examples would be rejected instead of silently accepted.
Outcome:
Created a structured eval/data pipeline useful for building high-quality AI training and evaluation datasets.
Tech:
OpenAI, Python, LLM Evals, Synthetic Data, JSONL, Agentic Workflows👍
0
21