I majored in Computer Science with a minor in Mathematics

My course of study was eclectic, as I continued to hold an interest in Economics, Sociology, and Literature. Ultimately however, Computing and Mathematics proved more satisfying. I’ve described my work as “solving Sudoku puzzles as I write them,” and making a rewarding career out of something I so deeply enjoy is truly a privilege.

UVM is where I got my start with Python, C, C++, C#, and Java.

Highlights of UVM

• Winner of the 2016 UVM CS Fair
• Dean’s List 2017
• I was the Treasurer of the UVM Renewable Energy Network
• I was the President of the UVM Anime Club

I began my studies in Economics and Literature, but shifted towards Computer Science and Physics over the course of my Sophomore year. Hampshire College is a small liberal arts school in the Pioneer Valley here in Massachusetts, and it lacked a robust Computer Science department. Thus I transferred to UVM between my Sophomore and Junior years.

I would list highlights, but none are relevant to my work, unless you, my potential colleague or employer are interested in my tenure as the bassist of a grungy indie band.

My most recent position was as the Head of the Computing Team of the Chung Lab, Picower Institute, MIT.

The Chung Lab is primarily a neural microscopy lab, collecting petabytes of data of human and mouse brain samples. That end, the imaging and comprehension of the brain, requires the implementation of cutting edge chemistry (to stain the samples), photo-electrics (to design the microscopes for imaging), and computational techniques (to handle and understand the data gleaned).

The brain is a ruthlessly convoluted thing, and sometimes the process of understanding it feels like a quixotic endeavor. However, the Chung Lab, along with the rest of the Picower Institute, and other labs across the world, are working each day towards such noble goals as curing Alzheimer’s, safeguarding mental health, and understanding thought itself.

As has been the case since the field’s inception, computing is essential to all efforts of neuroscience. In our lab’s case, my team was tasked with developing and implementing methods of compressing the raw neural imaging data into a usable format, and in deriving information from that data (such as nuclei shapes and counts, axon and blood vessel tracing, etc.).

Leadership and Project Management

In my capacity as Team Lead I managed the work of two post docs and assisted the lab’s five graduate students with the computing aspects of their projects.

I tried act as a mentor for my graduate students, assisting them with planning and pseudo code as much as implementing the techniques themselves.

Research

I performed my own research, and provided programming assistance for other’s projects. I assisted a wide range of research efforts, improving our nuclei detection and brain slab alignment, among other methods which I will refrain from mentioning for propriety’s sake.

My personal efforts regarded developing a method of efficient Fiber Detection to be utilized by the lab going forwards. My sole regret regarding my time in the position was leaving before a methods paper on the subject could materialize, but I’ve detailed what I can on this site in the link above. The method is being utilized in a collaborative paper between the Chung Lab and the Tsai Lab at the Picower Institute, for the NIH’s U01 Alzheimer’s Initiative, which is unfortunately why I can’t share the code directly.

Process Improvement and Automation

Given my skill set, I made an immediate effort to improve the lab’s processes, introducing automation, task scheduling, and quality assurance to our workflow. See the YAMOP project for more details, specifically the work I contributed to our paper in the journal Science.

As stated, I introduced task scheduling, resource allocation, streamlined quality assurance, and script batching to the lab’s workflow, primarily using bash and python scripts, though I utilized a fair amount of javascript for a browser based resource monitor.

When I arrived at the lab our graduate students were tasked with about an hour’s worth of computing work each day: running specific scripts, cleaning up old files, etc. There was no order to it, and entire computing servers would hang from too many concurrent processes, or worse, one process which error-ed out ungracefully and hogged resources.

When I left the lab, all computing processes associated with our regular workflow could be completed by a single staff member writing simple .json config files, and our graduate students could devote themselves to their actual projects. Tasks are scheduled and monitored, with any failures reported to a team member via a slack bot, and are distributed between our computing servers depending on load.

System Administration

A painful, but extremely necessary aspect of my position was the maintenance and administration of the lab’s sizable computing infrastructure.

That meant maintaining over 4 PB of storage spread across two distributed data clusters (one running CEPHFS and the other BeeGFS), over 600 TB in 12 separate ZFS data stores, five separate high powered computing servers, and an additional 500TB cloud storage container hosted on aws. I even ran and updated the lab’s apache II server, which served specific files and neuroglancer links to the public (an example of such a link I’ve added here.)

System Architecture

As part of our submission for the NIH’s UM1 grant program, I designed two new CEPHFS Data Clusters and five additional Computing Servers. Eventually we decided to move away from CEPH towards a ZFS solution running TrueNAS, which I worked with a colleague to design.

As part of the UM1 proposal we developed an entire data pipeline, from collection at the microscope, through multiple rounds of processing and analysis, to archiving on magnetic tape, and upload to a cloud storage server hosted on AWS.

Grant Writing

A surprisingly large portion of my work each autumn was spent grant writing. I’m quite proud to have written the computing portions of our successful UM1 grant proposal, awarding the lab millions of dollars to continue our work.

That being said, and being candid in the comfort of my own personal website, it was one of the most grueling processes I’ve ever encountered, far worse than system administration, or the more challenging (but far more rewarding) task of designing the proposed system itself. It may well have dissuaded me from remaining in academia in a managerial capacity.

Highlights of the Chung Lab

• Wrote Portions of Multimillion Dollar Grant
• Utilized Cutting Edge Techniques on Incomprehensibly Large Datasets
• Contributed to Building an Atlas of the Human Brain
• Gained Invaluable Experience in Systems Architecture, Admin, and Database Design
• Contributed to Discoveries in Alzheimer’s Research

I worked as a Data Operations Engineer, and was eventually promoted to Senior Data Operations Engineer after a series of unexpected departures left me the only Operator at the company for one frantic summer.

Panalgo is a data delivery company. We developed and maintain a web based application which allows policy makers and data scientists access to truly vast amounts of identifier scrubbed health insurance claims data. I’m proud to say that our services were used by policy makers and drug companies during the COVID 19 pandemic and, though my efforts were infinitesimally distant and diluted, I “did my part” to help during that crisis.

As for Data Operations itself, the position was a “meta-role,” so to speak, somewhere between a Software Engineer, Systems Architect, and Systems Administrator. We were responsible for creating and maintaining the backend systems which received data from providers, for running that data through the ETLs designed by our Data Engineering team, and making sure that the transformed data became available to clients.

Suffice to say, a great deal of our work was in Bash and Ruby, and I gained a true appreciation for interpreted programming languages, as well as the incredible breadth of packages available for Unix machines. QA and Automation were of the utmost importance, and my work evolved into the Yaeop (Yet Another ETL Operations Pipeline) project over the course of my time there (many aspects of which I later utilized in the Chung Lab’s data pipeline).

As I mentioned, after my first Summer with the company our team shrank from three operators to one, and I received a “battlefield promotion” so to speak. Over the following year I became the team lead and hired new Operators, as well as worked with our engineering team to improve and automate our systems.

I took to management very well, and found a great deal of joy in assisting my colleagues in their development and projects, even if the added responsibility for product delivery created an additional layer of stress.

Highlights of Panalgo

• Work was Directly Beneficial for a Global Problem
• Gained Valuable Systems Knowledge and Leadership Experience
• Gained AWS and Azure Experience through Panalgo’s use of Cloud Computing
• Developed YAEOP with Engineering Team to Automate Workflow

I worked as an ETL Engineer using PL/SQL for database management and the Informatica proprietary software to develop the actual ETLs.

While the majority of my work was dedicated to standard ETL Development, I was also tasked with general process improvements and the implementation of new technologies.

Tufts Health Plan is a health insurance company, and while it’s claims database is SQL readable, when I worked there the backend was still written in Cobol. For obvious reasons, it was a high priority to begin transferring the data to a new framework. While I can’t say whether the transfer process was actually completed, as I left before plans were finalized, I was able to investigate ElasticSearch and a couple other non-relational database frameworks as a potential replacement.

Highlights of Tufts Health Plan

• Developed Python Scripts to Automate Workflow (it has truly been a theme of my career)
• Gained Experience with ElasticSearch and Logstash, as well as SQL, Java, and the Informatica program

I worked as a lab assistant to Professor Joshua Bongard, a friend and a mentor. My time at the Complex Systems Lab was spent running experiments and updating a former graduate student’s project to a new physics engine. Living in Burlington Vermont in the summer and autumn, spending my mornings coding in a cafe or the lab, only to ride my bike down to North Beach on Lake Champlain, truly the magic of youth! Winter in Vermont was another story, unfortunately.

Highlights of the Vermont Complex Systems Lab

• Updated the “Twitch Plays Robotics” project to use the Unity Game Engine, running C# for both physics and web integration.
• Coded evolutionary algorithms for the ludobots project.

I worked as a lab assistant in the wet lab at the Mass General Endocrine Unit. My time was spent mixing reagents, pipetting, processing sacrifices (though I’m grateful to have only dealt with posthumous tissue, as opposed to having to perform the act myself), waiting for experiments to finish, and recording initial observations before the lab tech I was assigned to performed a more thorough analysis.

My time at the endocrine unit was formative, as it was the first time I glimpsed the messiness of the scientific process at work. Still, I admit the primary lesson I’ve kept into my current career (given the dissimilarity of skill set), is an abiding empathy for wet lab technicians and medical professionals of all sorts.

One shouldn’t need a thorough understanding to sympathize, but there’s a breadth of empathy only accessible to those who have been in the exact same shoes. I don’t mean to say that wet lab work is an excruciating nightmare, only that I appreciate people who get their hands dirty, having done so myself.