Ashu

AI/ML Engineer

Artificial Intelligence and Machine Learning
1roles
38skills
1education

About

AI/ML Engineer with 3.3+ years of experience designing and deploying production AI systems across computer vision, biometric verification, NLP, and Generative AI.

Experience

AI/ML Engineer

Cubastion Consulting Pvt. Ltd.

2023 - Present

Worked on the SSC Transformation Project, architecting and deploying a self-hosted Generative AI chatbot and agentic AI pipeline using open-source LLMs, LangChain, LangGraph, FAISS, prompt engineering, and secure document retrieval. Developed real-time computer vision systems for photo quality assessment, identity verification, candidate deduplication, video and facial emotion analysis, and a Flask REST API. Designed a cryptographic QR-based secure exam paper system for government recruitment workflows. Delivered systems achieving 99%+ accuracy for photo quality assessment and 98%+ accuracy for photo comparison at government and national examination scale.

Education

NSUT East Campus

B.Tech, Electronics and Communication Engineering

2018 - 2022

GPA: 8.26

Skills

PythonLangChainLangGraphModel Context Protocol (MCP)Prompt EngineeringRAG ArchitectureOpen-Source LLMsMistralLLaMAOllamaHuggingFace TransformersFAISSChromaDBEmbedding PipelinesMetadata FilteringTensorFlowScikit-learnYOLOModel Fine-TuningImage ClassificationLiveness DetectionBiometric VerificationOpenCVFace RecognitionReal-Time Video ProcessingImage Quality AssessmentEmotion AnalysisGitGitHubFlaskDjangoPandasNumPyMatplotlibPostgreSQLREST API DesignOn-Premise AI DeploymentLinux-Based Server Environments

Projects

Self-Hosted RAG Chatbot & Agentic AI Pipeline

Architected and deployed a fully self-hosted Generative AI chatbot using open-source LLMs, LangChain, LangGraph, FAISS, secure document ingestion, embedding generation, metadata filtering, and prompt engineering.

Live Photo Capture - Computer Vision Quality Assessment

Developed a real-time photo verification system using YOLO-based object detection and custom face-quality scoring heuristics to prevent captures with hats, masks, and glasses, achieving 99%+ accuracy.

Photo Comparison Engine - 1:1, 1:N, N:N Matching

Designed and deployed a high-throughput photo comparison system for identity verification, candidate deduplication, and batch matching, achieving 98%+ accuracy with results delivered in seconds.

Video & Facial Emotion Analysis API

Engineered a Flask REST API for face detection, frame-by-frame emotion classification, and structured storage from video streams and static images.

QR-Based Secure Exam Paper System

Designed a cryptographic QR encoding and scanning system for tamper-evident SSC question paper retrieval and exam confidentiality.