Machine Learning




You’ll be amazed by what’s become possible in the last three years.

Hi, I’m Lawrence McDaniel and I’m a technology freelancer. The rapid evolution of IT infrastructure cloud services combined with the sudden and staggering accumulation of electronic data has catalyzed an explosion of applied science and innovation for many ideas that only a few years ago were still not much more than theory. The math hasn’t gotten any easier, and IT infrastructure has actually become a lot more complex, however, If you understand both of these disciplines and you also are a halfway decent software engineer then the sky is the limit in terms of what you can do.

Fortunately some of the greatest minds on earth, like Dr. Andrew Ng at Stanford University for example, have paved the way for engineers like me to implement incredible machine learning and artificial intelligence systems that leverage big data sets to attack problem spaces like computer vision, pattern recognition in highly dynamic environments (like debt, equities and commodities secondary markets for example), big data classification and insight problems, and robotics.

Sample Code

Follow the link below if you are looking for code samples for the programming exercises for Coursera Machine Learning or Stanford University CS229 Machine Learning. This GitHub repository contains my fully-vectorized documented code implementations written in Octave. I took the course on Coursera, which I HIGHLY recommend by the way.


Designing infrastructure environments for machine learning projects is quite different from most web services and apps. High data throughput and horizontal scalability are critical. Moreover, providers like AWS are constantly evolving and introducing new services for extremely large data sets, analytics and map-reduce parallelism.


Designing data sets for machine learning projects is rapidly evolving into its own trade craft. Beyond traditional data QC and analysis and database skills you also need intuition into how algorithms work (or don't work) and how changing/increasing learning and cross-validation sets can move a project forward.


The good news is that few if any of the algorithms currently in use are new. The bad news is that understanding when to use each and their respective strengths and weaknesses is an ongoing and fast-evolving process.




Multi-variable Linear, Polynomial and Logistic regression models leveraging a plethora of cost optimization strategies


Neural Networks

For image processing, compression, robotics, OCR and other non-linear problem spaces.


Support Vector Machines

For classification problems in non-linear environments like text categorization, image sorting and hand-written character recognition


Anomaly Detection

Supervised learning algorithms to identify potential service failures, factory defects and other types of outliers in highly dynamic and non-linear environments.


Recommender Systems

Content-based recommendation engines e-commerce optimizations and other predictive customer recommendations


Unsupervised Learning

Clustering and K-means algorithms for applications like market segmentation, social network analysis and IT infrastructure management

  • MySQL, MS-SQL, Transact-SQL, MongoDB, Hadoop 100%
  • CloudFormation, Bash, Ansible, Jenkins, Chef, Puppet 90%
  • Wordpress, Django, ASP .Net 85%
  • English (native speaker) 100%
  • Octave, MatLab, MS Excel, MS Excel VBA 95%
  • Python, PHP, VBA, VB .Net, ANSI C, C++ 85%
  • Objective-C, Swift, Java 80%
  • Spanish (professional proficiency) 90%


Financial forecasting, fin-tech, and algorithmic solutions projects