Farhan A.

AI safety and safeguards for real world harms.

Helping founders build tools that support human agency by increasing privacy, security, and information integrity.

Helping service orgs build AI native products through the OpenAI x Agency fund AI4GD accelerator.

Exploring products to enhance user agency over their lives. aka how to use AI/LLMs to increase access to critical services (healthcare, govt services, financial mobility).

I led the Public Health Engineering teams at Meta, delivering the company goal of bringing 50M people one step closer to COVID-19 vaccine. Worked across recommendations (newsfeed, stories, reels), topic modeling, government engagement, civic and political content, blood donations, vaccine finder, local alerting, among others.

I led Meta: the flagship machine learning platform within Pure1 - covering product and eng, as we grew from a small idea with a few people to launch and subsequently the industry leading predictive analytics service for storage.

Worked on a variety of things ranging over - I led a team of engineers in a multi-year effort to build a distributed sequencing system for guaranteeing data resiliency in the high performance log-structured metadata database which drives I/O in our core system. - Implementing QoS algorithms for metadata fairness while writing to flash. - becoming intimately familiar with our database system and developed enhancements and critical fixes for the product. -Implementing cryptographic algorithms (HMAC, AES-KW) and their use for data at rest encryption to work towards getting a FIPS level 2 certificate for our cryptographic module from NIST

Worked in the Storage and File Systems team to develop real-time monitoring solutions for multiple tiered (SSD, HDD) systems of storage. Designed and implemented additions to two real-time monitoring tools - Server Performance Advisor and Perfmon, and published a white paper to customers explaining the new technology.

Organise two of our flagship events - Princeton Pitch and TigerLaunch.

Organized and worked as a TA in classes on basic and advanced web development and iOS programming.

Developed a project for using SMS texts from community agents (~20,000 agents) to gather reports and monitor health conditions in villages around Madagascar. Worked through the development, testing and deployment stages of the project and trained a pilot group of agents in the usage of the system. Programmed the system in Python using the Django and RapidSMS framework.

Best Thesis in Computer Science. Subsequently published at AISTATS