Ramin01 — EngineerSharifi
Transforming cutting-edge AI concepts into production-ready solutions. Building intelligent systems at the intersection of ML, software engineering, and autonomous agents.
About
Senior Machine Learning Engineer with deep expertise in time-series modeling, IoT data pipelines, and applied deep learning. I specialize in transforming complex AI research into production-grade systems that operate at scale.
My work spans sequential prediction using LSTM/GRU architectures, anomaly detection for sensor-driven systems, and large-scale data engineering pipelines deployed on cloud platforms. I've published in IEEE conferences and led teams building multi-agent AI systems.
Experience
- —Delivery manager and team lead for Google's EDU TerminalBench and Google's Apps Script Bench
- —Awarded Turing Google Hero of the Month for outstanding contributions
- —Designed real-time data processing pipelines for anomaly detection and forecasting in large-scale autonomous systems
- —Built scalable backend services using FastAPI, Celery, and PostgreSQL for high-frequency data ingestion
- —Architected multi-agent AI systems integrating predictive signals and cross-agent coordination
- —Developed full-stack dashboards in React.js/Next.js for visualizing time-series telemetry and live metrics
- —Led cross-functional engineering efforts integrating ML-powered analytics into production systems
- —Built end-to-end IoT data pipelines for high-frequency sensor streams using AWS/GCP
- —Designed ML models for sequential pattern analysis, forecasting, and anomaly detection
- —Optimized ML inference pipelines for near-real-time operation
- —Utilized YOLO V7 for object detection with automated annotation workflows
- —Integrated Hidden Markov Models into ML pipelines for sequential data analysis
- —Applied data mining techniques for business intelligence at scale
- —Conducted research in capsule networks for image classification and object detection
- —Optimized models using pruning techniques, reducing computational costs
- —Developed iOS apps with on-device ML inference using Swift and CoreML
- —Designed the Gelderm classifier for improved medical image classification
Tech Stack
Publications
Selected Work
Agentic AI Playground
Modular AI agents using Dify, LangChain, and CrewAI for multi-step reasoning and workflow automation. Integrated observability, vector search, and dynamic tool orchestration.
LangChain & CrewAI Pipelines
Reusable agent pipelines for software engineering automation — code refactoring, repository organization, and documentation synthesis.
Real-Time IoT Telemetry Dashboard
End-to-end pipeline for processing high-frequency sensor streams with live visualization of time-series data, anomaly detection alerts, and predictive analytics.
Gelderm Classifier
Mobile-optimized computer vision system for medical image classification. Separated detection and classification phases for improved accuracy with on-device inference via CoreML.
LSTM Activity Predictor
LSTM-based forecasting model for mobile user-activity prediction. Published at IEEE PACRIM 2019 — demonstrates sequential pattern analysis for real-world mobile data.