Protein Engineering Intern on the biohybrid BCI team. Screened ~20,000 KR2 opsin variants on Galago, generating libraries via error-prone PCR, assembly, and transformation. Automated colony picking across 25 384-well plates on a Hamilton liquid handler and built a Nanopore pipeline for library sequencing. Designed and verified dual-expression opsin plasmids enabling parallel characterization in E. coli and mammalian cell lines.

SDE Intern on the Devices team. Developed and optimized C++ middleware services powering Alexa voice experiences across Fire TV devices, reducing memory usage and improving runtime reliability. Wrote technical design documents and drove cross-team integration for a C++ feature spanning multiple Fire TV services and interfaces.

Built a calorie detection app by integrating smart glasses image capture, cloud-based calorie estimation, and frontend UI. Tested computer vision models including YOLOv8, SAM, and Depth Anything, and worked with vendors on camera requirements, cost, and hardware constraints.

Helped build AI-scientist systems to automate life science research, including bacterial transformation, Nanopore sequencing, colony picking, and custom lab automation tools.

Built Prophet-based forecasting for 15-week Mac sales, improved MAPE by 56%, and incorporated the final model into the data team's prediction workflow.

Cogan Lab

Developed end-to-end MNE-Python pipelines for intracranial sEEG analysis, from Audacity response annotation and high-gamma extraction to permutation-based electrode visualization for speech perception decoding. Analyzed 26 patient datasets to identify phoneme-dependent uniqueness points and characterize inter-subject variability in high-gamma neural representations.

Chory Lab

Engineered automated T7 phage-directed evolution workflows on Hamilton liquid handlers for 72-hour autonomous operation. Reduced PCR and Qubit reaction cost by over 50% on the Opentrons OT-2, contributed PyLabRobot visualizer modules, and documented automation protocols for future use.

Barrett Lab

Conducted Cell Painting Assay on induced iPSC-derived neurons and progenitor cells to study Down Syndrome morphology. Built an XGBoost pipeline to rank discriminative imaging features, achieving 83% accuracy in classifying Down Syndrome versus control cells.

Musah Lab

Analyzed RNA-seq data from iPSC-derived podocytes under injury conditions to identify potential therapeutic targets for chronic kidney disease. Gained hands-on experience with stem cell culture, injury treatment, and organ-on-chip technology; second-author manuscript under review at Stem Cell Reports.

Navid Lab

Designed a primal-dual constraint learning framework for protein pathogenicity and fitness score prediction on ProteinNPT data. Fine-tuned ESM2 8M with LoRA on about 700k variants, improved Spearman correlation by 2% over baselines, and built a scalable 5-fold PyTorch validation pipeline.