Electrical & Computer Engineering + Computer Science — Duke
Evan Merzon
From custom silicon to production code.
Splitting time between Rubrik, MetLife Investment Management, and Oscilink — the BCI startup I co-founded and lead as CTO. Earlier: built a custom processor in Verilog for an FPGA DJ board, and trained CNNs on pathology slides at Mount Sinai.
B.S.E. Electrical & Computer Engineering · B.S. Computer Science
GPA 4.0 — Pratt School of Engineering Dean's List with Distinction, every term
Software Engineering Intern
- Migrated AKS cluster infrastructure from Cluster Autoscaler (CAS) to Node Autoprovisioning (NAP)
- Built out spot-aware scheduling alongside NAP, cutting compute costs 53% in a proof-of-concept
- Migrated cluster networking policy enforcement from Calico to Cilium
- Built a tool to analyze slow queries across the team's Azure Postgres instances
Software Engineer · Part-time (Intern May – Aug 2025)
- Deployed a local LLM on a RAG pipeline with spreadsheet agents to build a chatbot for querying hundreds of financial documents and spreadsheets
- Built and productionized a client-facing investment portal, projected to save $2M annually versus the outsourced alternative
- Built an internal plotting/reporting app for the firm's quant investing model — key to securing $350M+ in AUM, still used weekly for client reporting
- Leading efforts to bring the portal to clients
AI Summer Intern
- Independently developed and deployed a whole-slide image viewer for digital pathology, migrating it from Docker to the hospital's Singularity servers
- Automated integration of Mount Sinai's multi-petabyte pathology slide repository (1M+ slides, 2PB+) with the viewer, including nightly no-duplicate ingestion scripts and MongoDB backup automation
- Added AI heatmap overlays so users could visualize predictive model results directly in the browser
- Rehired in 2024 to train a CNN for nuclei centroid detection matching state-of-the-art supervised models, and optimized inference for efficient annotation of multi-gigabyte images
Research Intern
- Used MATLAB and Python for MRI image analysis and motion correction
- Implemented automatic coregistration and white matter mask generation for semi-automated segmentation of perivascular spaces
- Trained a CNN for motion artifact correction as a proof-of-concept for the lab
Research Assistant
- Developed and evaluated a CNN for segmentation of vagus nerve histology slides
- Performed human-in-the-loop training of myelinated fiber segmentation models and quantified metrics like fiber dimension and count
- Automated QC across millions of nerve fiber segmentations in thousands of images from 100 cadavers, boosting model accuracy 10%
Junior Software Engineer
- Delivered software used by 50,000+ seniors and staff across 600 nursing home facilities, shipping features to Celo Engage, Celo Context, and ComPASS
- Built a 15+ stage MongoDB aggregation pipeline joined with MySQL for a HIPAA-compliant analytics page, with caching for repeat queries
- Implemented an AWS Lambda job to automate cleanup of unused customer devices and send inactivity warning emails, plus staff-facing widgets summarizing resident engagement
- Wrote Selenium test coverage for new endpoints and features, including a virtual coloring book and a Wordle clone for residents
- Merged 800+ pull requests, reviewed 1,500+
Startup — Hardware + ML
Oscilink
Co-Founder & CTO
Building EEG electrodes with higher spatial resolution and cleaner signal to power next-generation brain-computer interface applications.
- Leading the ML processing pipeline that turns raw electrode signal into usable, high-fidelity data
- Building the data acquisition pipeline connecting hardware to software
- Co-founder and CTO — hardware, ML, and product direction
Course Project — Digital Systems
DJ Board on FPGA
Independent build, Mar – May 2026
A fully custom audio computer built from the ground up — own processor, own memory interface, own board.
- Designed a custom processor from scratch in Verilog
- Wrote real-time SD-to-DDR2 audio playback logic in Verilog and Assembly running on that processor, plus a VGA UI
- First project in 25+ years of the course to get the board's onboard DDR2 memory working — no prior code or notes existed to build on, just a barely-documented memory controller
- Debugged the memory interface using the board's 16 onboard LEDs as the only output, then documented the process for future students
Department Tooling
PrairieLearn Intake Assessments
Duke ECE Department
- Built intake assessments for the Duke ECE department on PrairieLearn, the open-source platform Duke uses for auto-graded coursework
- Authored questions and workflows used to place and evaluate incoming students
Team Project — Duke Aviators
Flight Simulator Backing
Structural Build
- Designed and built the wood, laser-cut, and 3D-printed backing structure that houses and conceals the simulator's MFD/PFD displays and connector panel
- Built as part of a team through Duke Aviators
- Cable management was the hardest part of the build — routing HDMI, USB, and power cleanly through the enclosure for two displays
Languages
Hardware
ML / AI
Frameworks & Tools
Cloud & Data
Developed GoBringIt, a mobile food-ordering app for Duke students on food plans, used daily by hundreds of students. Built real-time order tracking, promo codes, and in-app credit.
Co-founder of a club partnering with several VC firms to give Duke students real-world sourcing and due-diligence experience. Helped select 30 students from 150 applicants.