Meta Superintelligence Labs (MSL):

(Building the infrastructure to power Personal Superintelligence)

• Key member of the Core Engineering team driving the rollout of Metamate 2.0 which is Meta's flagship Internal AI Platform and Agentic Coding Tool. This tool's progress is company-level priority of C-suite executives.

• Designing and scaling the foundational infrastructure required to support next-generation AI capabilities.

• Specializing in Distributed Systems, Performance Optimization, and Telemetry/Observability (OpenTelemetry/OTEL) to monitor and optimize critical metrics, including tool calls and Time To First Token (TTFT).

PREVIOUS:

Network Planning Systems Engineering (NPSE):

(Developing advanced systems to model and scale Meta's global backbone network)

• Optimized Meta's backbone network design by developing predictive models leveraging a combination of Integer Linear Programming (ILP) Solvers and Reinforcement Learning.

• Built reliable, large-scale distributed systems from the ground up utilizing C++, Rust, and Python.

• Collaborated extensively across teams to forecast network demand, build infrastructure capacity, and mitigate risks within Meta’s production network—directly supporting the modeling and planning of new Data Centers, including dedicated AI clusters.

Worked on the RobotPt Cordial project in collaboration with UCLA. This project involved the development of robots to assist people with low-vision, and my ability to learn new skills quickly allowed me to contribute to the design and implementation of innovative solutions that could improve the quality of life for individuals with visual impairments.

I have successfully developed an end-to-end DCRNN machine learning model for predicting road traffic in Los Angeles, using real sensor data to train the model. I have also created a user-friendly front-end using ReactJS that incorporates the machine learning model into a Python/Flask backend, enabling users to plan the best route from point A to point B within Los Angeles and compare the results to other models like HA (Historical Average), FC-LSTM, and ground truth.

Designed and developed a message queue architecture for distributed computing. This will be used to solve optimization problems of Xpress and Gurobi solvers to predict the next network demand in Facebook Networks to plan resource allocation for the future.

As a ML Intern, I have had the opportunity to use these technologies to develop solutions for vehicle speed estimation and the detection of over-speeding and dangerous lane changes. My expertise in these areas allowed me to create algorithms and predictive models that could accurately estimate vehicle speed and identify potential safety hazards on the road. I was responsible for the full development lifecycle of these solutions, from initial concept to final implementation and deployment. Through this project, I gained valuable experience in applying advanced technologies to solve real-world problems, and pushing the boundaries of what is possible with Computer Vision and Machine Learning.