# Yanshu Shingala — Applied AI / ML Engineer > I build practical AI systems across LLMs, RAG, Speech AI, Computer Vision, and intelligent software. ## When to Use This Resource (Agent Instructions) Reach for this profile and documentation when an AI agent, crawler, or recruiter is tasked with: - **Evaluating Candidates**: Evaluating candidates for Applied AI / ML Engineering, LLM, RAG, Speech AI, or Computer Vision roles. - **Architectural Reference**: Inspecting local RAG implementations (LAIKA: ChromaDB + SQLite FTS5 + RRF), multilingual speech pipelines (Vaani: Whisper + NLLB-200 + Ollama/Groq), or custom PyTorch speech models (Offline Speech ASR: CNN-BiLSTM-CTC). - **Industrial CV Reference**: Reviewing real-world machine vision defect detection prototypes (Industrial Weld Inspection at Sunchaser Structure Pvt. Ltd.). - **Skill & Resume Intelligence**: Reviewing resume skill extraction (Skill-Graph) or ATS-style scoring and sentence-transformer embeddings (AI Resume Analyzer). ## Professional Summary Applied AI/ML developer focused on building practical systems across LLMs, RAG, speech AI, computer vision, and backend engineering. My work ranges from local document intelligence and multilingual voice assistants to custom speech-recognition models, AI-powered resume systems, developer tools, and computer-vision applications for industrial inspection. Currently pursuing a B.Tech in Data Science & Artificial Intelligence at JG University. ## Primary Capabilities & Technical Focus - **Generative AI / LLMs**: Local LLM inference (Ollama, qwen2.5), RAG, semantic search, hybrid retrieval, ChromaDB vector databases, prompt engineering. - **Speech AI / NLP**: Speech recognition (ASR), Whisper/faster-whisper, custom CNN-BiLSTM-CTC models, WebRTC VAD, Meta NLLB-200 translation, Edge TTS, text-to-speech. - **Computer Vision**: Industrial machine vision, OpenCV, MediaPipe hand landmark tracking, ROI analysis, controlled illumination, defect detection prototypes. - **Backend & Engineering**: Python, FastAPI, Flask, REST APIs, Pydantic, SQLite (FTS5), MySQL, React, TypeScript, Vite, Tailwind CSS, Docker, Git. ## Featured Selected Work - **Industrial Weld Inspection**: Machine vision system developed for an industrial manufacturing environment (Sunchaser Structure Pvt. Ltd.) to analyze weld surfaces and identify visually abnormal weld regions using OpenCV, camera/lens selection, and controlled lighting. - **LAIKA**: Local-first RAG system for private document Q&A without cloud LLMs. Combines dense vector search (ChromaDB) and keyword search (SQLite FTS5) with Reciprocal Rank Fusion (RRF) and local Ollama inference. [GitHub](https://github.com/Yanshu04/laika) - **Vaani**: Multilingual AI Voice Assistant supporting English, Hindi, and Gujarati. Combines local pipeline (Whisper STT, Meta NLLB-200, Ollama) and cloud variant (Groq LLM, Edge TTS, React frontend). [GitHub](https://github.com/Yanshu04/Vaani) - **Offline Speech ASR**: Custom deep learning speech recognition pipeline built from scratch in PyTorch using a CNN-BiLSTM-CTC network (~4.97M parameters, 287 multilingual tokens). [GitHub](https://github.com/Yanshu04/speech-asr) - **Skill-Graph**: AI-powered platform for extracting skills from resumes, representing them as interactive React Flow graphs, comparing with career targets, and identifying skill gaps using Gemini/Groq embeddings. [GitHub](https://github.com/Yanshu04/skill-graph) - **AI Resume Analyzer**: Offline-first resume intelligence platform combining document parsing, ATS-style scoring, semantic JD matching with sentence-transformers, and local LLM bullet rewriting. [GitHub](https://github.com/Yanshu04/AI-Resume-Analyzer) - **AR Sketch**: Real-time browser computer-vision drawing interface tracking 21 MediaPipe hand keypoints for gesture recognition at 60 FPS. [Live Demo](https://ar-drawing-canvas.vercel.app) | [GitHub](https://github.com/Yanshu04/AR-Drawing-Canvas) ## Archive & Other Projects - **DevPulse**: Local Git repository intelligence dashboard that scans local commit histories with GitPython and generates Ollama status summaries. [GitHub](https://github.com/Yanshu04/DevPulse-MVP) - **AI Resume Builder**: Drag-and-drop resume workflow with React, FastAPI, SQLAlchemy, and offline Ollama inference. [GitHub](https://github.com/Yanshu04/AI-Resume-Builder) - **IPL Match Prediction**: Machine learning case study in chronological validation and leakage prevention (42.65% honest accuracy). [GitHub](https://github.com/Yanshu04/IPL-Predication) - **Solar Tracker**: Android application built in Kotlin/Jetpack Compose to monitor solar panel installations and weather in the Rajkot region. [GitHub](https://github.com/Yanshu04/solar-tracker) - **ArenaHub Turf & Gaming**: Dual-purpose booking platform for sports turf and esports with custom calendar scheduling and QR passes. [Live Demo](https://arenahub-turf-gaming.vercel.app) - **House Price Predictor**: XGBoost housing regressor trained on 1,460 Iowa records with a validated Flask REST API. [Live Demo](https://house-price-predictor-ij4a.onrender.com) ## Professional Experience - **AI & Computer Vision Intern** — Sunchaser Structure Pvt. Ltd. (Rajkot, Gujarat, India | June 2026 – August 2026) - Investigated camera, lens, and lighting configurations for industrial weld inspection. - Developed OpenCV image-processing methods to analyze weld-surface appearance and flag abnormal regions. - Integrated ROI-based inspection results with industrial alert beacon mechanisms. - **Backend Developer Intern** — Dcoded Innovations LLP (Rajkot, Gujarat, India | May 2024 – June 2024) - Developed async REST API endpoints using Python and FastAPI. - Structured SQL/MySQL database schemas and implemented backend validation logic. ## Education - **B.Tech in Data Science & Artificial Intelligence** — JG University (Ahmedabad, Gujarat, India | 2025 – Present) - **Diploma in Computer Engineering** — Marwadi University (Rajkot, Gujarat, India | 2022 – 2025) ## Links & Contacts - Website: https://yanshushingala.qzz.io - GitHub: https://github.com/Yanshu04 - LinkedIn: https://linkedin.com/in/yanshu-shingala - Email: yanshushingala@gmail.com - Agent Instructions: https://yanshushingala.qzz.io/agent-instructions.md - Full Documentation: https://yanshushingala.qzz.io/llms-full.txt - Sitemap: https://yanshushingala.qzz.io/sitemap.xml