AI/ML
Engineer

An AI-powered platform for learning Tibetan, built at the Monlam Hackathon 2026. Three interconnected products — a vocabulary-grounded AI tutor, practice drills across four linguistic skills, and a fully autonomous bilingual newsroom.

• Sherab Tutor: grounds every definition in real dictionary sources before the LLM explains — no invented vocabulary
• Drills: listening (TTS in Lhasa, Amdo & Kham dialects), speaking (STT), stroke-order tracing, glyph assembly
• Agentic newsroom: crawls a Tibetan news corpus and autonomously writes a weekly bilingual issue — no human editor

An automation tool that streamlines the job hunt: it searches postings, filters them against your criteria, and submits applications on your behalf.

• Automated job search and application submission
• Smart filtering based on job criteria
• Résumé and cover-letter customization

A RAG-based virtual assistant built on GPT-4 that handles reservations and orders with the efficiency of a machine and the courtesy of your best staff member.

• Handles multiple concurrent customer requests
• Consistent, accurate, grounded responses
• Retrieval-augmented generation over menu & policy data

A deep learning model that detects mechanical faults in heavy-industry rotatory machines from their vibration signals.

• 98% accuracy in fault detection
• Deployed at Machine Monitoring Systems, LLC

A computer-vision tool that extracts figures and their associated captions from PDF documents — built for assembling LLM training corpora.

• Detects figures, diagrams, and plots automatically
• Locates figure boundaries with CV techniques
• Pairs each figure with its corresponding caption

A data-collection tool for assembling large-scale text corpora used in training and fine-tuning large language models.

• Generates download links from a search term, across pages
• Downloads hundreds of books in parallel with error handling
• Fully automated pipeline from search to storage