Experience

Amazon

Will be joining Amazon as a Software Development Engineer Intern for Summer 2026, working in a large distributed environment with regular manager support and day-to-day engineering collaboration.

• Will work with experienced engineers to design and build software for large-scale distributed systems.
• Will help build scalable storage, indexing, query, or prediction-oriented systems depending on team scope.
• Will design solutions from broadly defined problems and ship them in an agile development environment.

NYU Tandon School of Engineering

Worked as a course assistant for Signals, Systems and Transforms under Prof. Ivan Selesnick, helping with labs, grading, quiz material, and office hours.

• Prepared model solutions and grading rubrics using MATLAB, `matplotlib`, and `scipy` for signal-processing topics.
• Reviewed 50+ submissions with rubric-based feedback on convolution, Z-transforms, frequency response, and elliptic filters.
• Designed multi-part quiz and practice material, then used office hours to help students move from formulas to workable problem-solving steps.

TREx Research Fellowship, NYU Tandon

Working on ML and computer-vision pipelines for high-resolution historic map analysis, with the goal of turning visual feature extraction into GIS-compatible spatial data that researchers can actually use.

• Building feature extraction, geotagging, and validation workflows for historic urban maps under a funded summer research fellowship.
• Using statistically grounded validation and map-specific CV pipelines so the output is usable beyond a one-off notebook experiment.

NYU VIP ARC Robotics

Worked on the perception pipeline for Team UltraViolet, NYU's VIP robotics team in the RoboMaster ecosystem. The work was less about "training a model once" and more about tracing where frame time was going in a ROS2 plus YOLOv8 stack, then deciding which parts belonged in Python and which parts should eventually move closer to TensorRT and C++ deployment.

• Mapped the full detection path from camera frame to preprocessing, GPU inference, post-processing, and publish, then used that analysis to isolate the hot spots worth fixing first.
• Reviewed the CPU letterbox path in utils.py and documented that resize plus padding cost about 2-3 ms per frame before inference, with an additional host-to-device copy immediately after preprocessing.
• Compared the existing Python rclpy node against a TensorRT-based C++ ROS2 branch and concluded that Python was better for iteration, while a C++ path with pre-allocated CUDA streams was the better final deployment target on Jetson.

NYU Secure Systems Lab

Worked within Justin Cappos's Secure Systems Lab on workload isolation and deployment questions for machine learning environments. This was systems-facing lab work centered on runtime boundaries, deployment assumptions, and how much trust ML tooling quietly expects from the host underneath it.

• Spent most of the work on Linux and virtualization boundaries rather than model code, especially where shared environments make "just run the notebook" a weak security assumption.
• Used that work to build stronger intuition for isolation tradeoffs, lab environment setup, and what has to change when an ML workload stops being a single trusted local process.

Conda

Contributed upstream changes to Conda around CLI behavior and dependency-resolution paths rather than keeping local fixes private to course or personal tooling.

• Merged upstream PRs improving CLI error handling and dependency resolution in the Conda package manager.
• Worked in the maintainers' review loop and treated the contribution like real ecosystem-facing engineering rather than a local patch.

SalesUp

Returned to SalesUp full time after interning there during my final undergraduate year. Worked on backend systems, AI-powered automation, and delivery workflows, with a lot of the work focused on cleaning up request paths and moving long-running operations into better async execution flows.

• Moved heavier work off synchronous request paths, which improved throughput by 40% and reduced P99 latency by 25%.
• Built and maintained FastAPI, PostgreSQL, Redis, Docker, and AWS-backed services.
• Worked on backend APIs, internal automation, and LLM-based product features with the surrounding CI/CD and deployment support needed to keep them usable.

SalesUp

Started at SalesUp as an intern during the final year of my undergraduate program, building backend APIs and internal automation tooling before converting into a full-time engineering role.