Terran Melconian

Data Science Consulting and Training

I provide advice and prototyping to companies considering a data science or analytics effort, and training programs to move software engineers and analysts into data science roles, which can be a very cost-effective addition to the recruiting pipeline. I would be pleased to discuss the potential benefits at your company.

I have made some conference presentations on these and related topics.

Advice to Prospective Learners

If you are an individual wanting to develop your own proficiency in the data science field, I recommend studying the following books:

Additional book reviews are available on Goodreads.

I advise you to study Data Science, a broad field, not Machine Learning, a narrow one. The latter, as it is currently framed, is valuable only at the doctoral level; it specifically covers algorithmic work to reduce errors in future predictions. You need to know about correctly running experiments, assessing business opportunities, developing an intuition for your problem, and determining importance and causality, which are the tasks that most businesses actually need done, not a mediocre reimplementation of image classification.

Research Interests

Operational Anomaly Detection
As software operations becomes increasing complex and increasingly dominated by undesigned emergent behavior, hand-coded alerting thresholds on a single metric are proving inadequate. The next generation of alerting will be multidimensional and learned. I am seeking to partner with companies which can release anonymized operational data under an open license to facilitate algorithm testing and development, who will then benefit from open-source algorithms tuned for their use cases.

Cryptocurrency Trading
The cryptocurrency markets interest me as an example of a less mature and less efficient market, presenting an opportunity for study whose like I was not alive to experience in equities.

Statistics for Personal Health
Experiment design becomes more challenging when there is exactly one subject available, yet this is exactly the situation we face in identifying food allergies and assessing individual medication and supplement effectiveness. How can we make more powerful and more usable tools, especially for use by people who may be cognitively impaired because of the very problem they are trying to diagnose?

I welcome correspondence regarding collaboration on these topics.


You can reach me at . If you can't see that address, you can take my first name, an at-sign, and then the domain you are currently viewing. I am usually able to reply to most inquiries from those seeking career advice.

Brief Professional Biography

Past Projects

Some interesting things I have done, of an avocational nature:

For Bots and Archivists

Many of my other projects are now obsolete and of interest to essentially no-one; they are here for search engines to crawl and to avoid breaking existing deep links: