About

Welcome.

 

My name is Ali Magine. I am a physicist by training, a data scientist and AI product builder by trade, and an entrepreneur by attempt! 

book

In my day job, I come up with innovative AI ideas and bring them to life quickly in enterprise applications. I direct data science and AI product teams, from conception to adoption, leveraging deep mathematical and algorithmic understanding in combination with an entrepreneurial mindset and experience. 

I have always been super passionate about better human thinking and the prospect of thinking machines that can make us collectively smarter. In my free time, I think deeply about foundations of intelligence. I have written a book to clarify the problem statement about “intelligence”, you know, the problem that the likes of Google Deepmind would like to solve.

Official Bio

Ali Magine holds a Ph.D. in theoretical physics from the University of Florida. In 2011, he (along with Dmitrii L. Maslov) predicted the existence of new emergent particles (relevant for quantum computing), evidence for which has been accumulating since 2015. He also published the main piece of his doctoral work with one of the most distinguished physicists in the field, Emmanuel Rashba, at Harvard University. Obsessed with investigating the question of how exactly our brain does physics, what are the principles of thought, and how they could be replicated in machines, left academia to work in the big data and AI industry since 2012. He has been ambitious in working with a number of startups (delivering high-impact AI solutions to a wide variety of industries), two of which got acquired by Intel corporation and Nokia. Formerly served as the lead scientist, lead inventor, and product owner of Intel’s cognitive computing platform. Co-Founded ii-AI in 2018, as Chief Innovation Officer, which turned into an AI consulting practice, and later joined Accenture where he is currently an AI lead in global AI solutions group.

Data and AI Experience in Industry (pre-2020)

 Online predictive and prescriptive optimizations 

Predictive maintenance, spare parts planning, reduce downtime
Hub and scheduling optimizations
Prediction and optimization of asset degradation
Reducing operating cost and improving safety
Predict valve and pump failure for data-driven resource/capital planning

Human Resources

Optimal staff hiring plan
Alternative plans for meeting deadlines
detecting overloaded experts, training opportunities, etc.

Healthcare

Surpassing cardiologists accuracy in highly similar conditions Automate echocardiogram diagnoses High performance bioinformatics in integrated ML-based LIMS

Issues and Defect Resolution

Surfacing similar issues across the organization, save time, labor and prevent defects
Reduce time to resolution by recommending applicable and similar resolutions

Internet of Things (IOT)

Fleet management, route analytics, cost optimization
Connected home use cases
Congestion management, blackout avoidance
Online asset survival analysis

Personalization

Product recommender systems
Help desk analytics
Detect customer state change

Customer Analytics

Know Your Customer (KYC)
Profiling
Anti-money laundering (AML)

Discovery and Root Cause Detection

Alias detection
Root cause detection of failure or defect

Defense

Surfacing new and interesting cases
Terror risk detection
Multimodal data unification for automated discovery

 

Anomaly and Fraud Detection

Insurance fraud: up-coding detection, fraud ring detection
Network intrusion detection

Capital planning and optimization

Investment policy optimization
Commodities trading optimization

Research

Research book on Rethinking Artificial Intelligence — Fundamental Intelligence

Journal publications – Google Scholar other research here

* Note a change in last name in 2019: Ashrafi —> Magine.

Patents

  • INTEGRATED INTELLIGENCE SYSTEMS AND PROCESSES (16/365,400)
  • SIMILARITY-BASED REASONING WITH ENSEMBLE MEMORY VOTING (15/201,102)
  • METHODS AND APPARATUS FOR DETECTING ANOMALIES IN ELECTRONIC DATA (15/392,837)
  • METHODS AND APPARATUS FOR IDENTIFYING NOVEL INFORMATION (15/425,670)
  • METHODS AND APPARATUS TO PREDICT SPORTS INJURIES (15/476,487)
  • EPISODIC MEMORY REPRESENTATION AND CONTROLLER FOR LEARNING (Intel filing)
  • UNIVERSAL QUERY ENGINE FOR REASONING PROCESSOR OVER EPISODIC MEMORIES (Intel filing)