Thomas Fuller Email & Phone Number

San Francisco, Ca, Us

San Francisco, Ca, Us

Menlo Park, Ca, Us

Building software to plan and operate Meta's AI zone and backbone networks.I build software components like:- Discrete-event network simulators for AI zone and backbone networks- Highly optimized, data-intensive, and horizontally-scaled C++ services handling millions of requests/day- Automated ML workload trace analyzers, enabling replay of real ML workloads in simulation environments- Experimentation frameworks for validating network simulator performance- Data pipelines for network infrastructure data to feed our simulators- Network risk management workflows and automationsThe tools I build answer these questions and more:- What is the expected network communication time for a given ML workload (e.g. llama3) on a AI zone?- How should we design our AI zones' network topologies?- How effective are our traffic engineering solutions at delivering AI zone network and backbone network performance?- If we need to perform maintenance on a given network device, what will the impact be on production backbone traffic or on ML workload performance in an AI zone?- How much time do we have before this datacenter runs out of network capacity?- How many concurrent network failures can we tolerate before dropping traffic?- What did the backbone network topology look like at a specific date and time?- If we are at risk of dropping traffic, where can we add capacity or reduce demand to mitigate the risk?I build tools for, and collaborate with, these people:- Network engineers- Network planners- Network demand forecasting specialists- Software engineers- Data scientists- Engineering managers and directors- Fiber optic delivery teamsThe tech stack I use:- C++- Python- SQL (Presto, MySQL, SparkSQL)- Thrift (RPC)- CI/CD- Data pipelines (like Airflow)- Jupyter Notebooks (pandas, matplotlib, scikit, etc.)- Software performance profilers

Interned on Meta's A/B Testing Framework Team. - Used machine learning to optimize queuing of requests to our data analytics service that processes petabytes of data each day. - As of February 2022, this feature is still in use and helps the service run smoothly to enable data scientists and others at Meta conduct feature experiments to continually improve Meta's products.- Landed improvements and features to a complex service running in production and actively utilized across Meta- Worked in a codebase that was in the process of migrating from python2 to python3, contributed to that effort as well as better engineering practices- Used Thrift RPC to communicate with other services at Meta- Collaborated with product managers, designers, data scientists, and engineering managers- Landed my internship's stretch goals and in my last weeks as an intern was able to contribute to other services the team was working as well

Seattle, Wa, Us

Interned on AWS' Training and Certification Team. - Supported and added features to a services backing AWS's training tool's learning management system - Implemented additional endpoints to improve the service's API compliance to the learning record store (LRS) standard known as xAPI (https://xapi.com)- Used serverless AWS services such as Dynamo and Lambda- Navigated changes to the team, including changing managers part way through the internship and losing many teammates- Demonstrated my intern project to the wider organization - Received a return internship offer

Think 'moneyball' but for rowing.- Compiled athlete performance data and practice logs into a centralized database- Used newly compiled database for dashboards and higher-level performance metrics- Designed and experiments to compare our team's training to other high-level rowing programs, such as the Dutch National Team's published workouts and results- This work was novel for rowing, and represented the union of my expertise as a coxswain and as a computer science student- In the years after my work, the team went on to win PAC-12 and national championships- I also had other responsibilities, such as inventorying the team's assets

Palo Alto, Ca, Us

Interned in a neurosurgery research laboratory.I developed and conducted an experiment using the resources and guidance available to me in the lab. My experiment explored the effects of concurrent treatment of Temozolomide and Ibuprofen on an immortalized glioblastoma cell line. I analyzed my data and organized them into an informal presentation. Additionally, I executed and aided in lab protocols for the laboratory's main experiments. These included, but were not limited to immunofluorescence stains, western blots, and assisting in work with lab animals. Growing and maintaining cell cultures using aseptic hood technique was a large component of my work. I also performed lab maintenance, cleaning, and related activities such as aliquoting, autoclaving, restocking, and washing.

Bachelor Of Science - Bs, Computer Science

High School Diploma