The Foundation DB Team.

Built a highly stable and efficient distributed file system in Golang with the following feature highlights: * Supports integrations with all major cloud storage providers: S3, Azure Blob Storage, and Google Cloud Storage. * Support large file operations (read, write, copy, unzip, etc.) at scale with O(1) memory usages by leveraging gRPC streaming and in-house designed worker pool. * Guarantees read-after-write consistency even for storages that have weak consistency (NFS/SMB) by leveraging request routing using Envoy. * Full support for RESTful API and RPC clients used by customers, automation processes and IB-owned applications. * Optimized both the latency and cloud cost by a file cache layer. * Readily to be deployed with Instabase's SaaS offering with SOC2 compliance; support highly customized storage life-cycle management policy by an in-house designed distributed crawler. Designed and implemented a system test framework from scratch to perform end-to-end API testing: * Covering major infra and product areas including the file system, flow execution and ML-studio. * Developed a full stress test suite for large-scale operations such as uploading large files (>2GB). * Integrate the system tests with the platform, allowing customers and site admins to run end-to-end tests on the platform to assess the platform health.Closely working with executives, support and sales teams to ensure customer success. Highlights: * Fully resolved a critical customer issue that impacts a multi-million dollar contract with a top-tier bank in a timely manner; the issue is multi-fold and is related to python's gevent limitation and thrift's connection pool design. * Extensive on-prem debugging for a customer and root-caused intermittent slowness in request handling to be at the network-level in customer's environment.Mentoring 10+ junior and senior engineers with design review, code review and hands-on debugging when necessary.

* Developing an in-house designed compiler to support a schema language (a.k.a. TACC) which generates C++ code for an event-driven framework that utilizes asynchronous I/O (e.g. epoll / inotify). Maintained a codebase with >1M lines of C/C++ code for both the TACC compiler and distributed runtime.* Re-architecting the TACC framework to support executing multiple agents in separate threads in parallel, which effectively improves resource usage and runtime performance. * Designed and implemented a new deletion protocol that unreferences an object upon deletion and defers its destruction in a distributed object-oriented framework. * Implemented a mount facility in TACC distributed runtime that synchronizes objects between agent and server, and handles disconnection and failover.* Implemented a debugging facility for coroutines that execute under the TACC distributed runtime. Refined the compiler to generate codes that compute the states of a coroutine in a stack-based form.* Developing a Fault Scenario Diagnosis (FSD) software based on TACC, which performs fast automatic root-cause analysis upon telemetry data obtained from a real-time system.* Bringing the FSD software to scale via performance analysis and profiling using tools such as callgrind and kcachegrind. Resolved critical performance bottleneck by improving the runtime performance by 100x and reduce memory consumption by two orders of magnitude. * Led the development of a scalability test suite with careful design and implementation. * Day-to-day mentoring junior engineers with hands-on support and guidance.

* Developing multi-physical models to understand why and how dose the human cornea swell under different situations (e.g. in vivo, in vitro, or pathological conditions)* Developed a fluid-structure interaction model to study the transient behavior of the human cornea under air-puff, for the purpose of improving the measurement of intraocular pressure (IOP) which is crucial for the detection of glaucoma* Invented a theoretical and computational framework to calculate the interacting forces between the collagen fibrils in the corneal stroma. The model proposed a new mechanism of the self-organization of the collagen fibril lattice. * Programming intensively in Matlab developing complicated models

ME335A-C, Finite Element Analysis. * Developed a Matlab framework for students to solve various types of mechanical problems including the heat-transfer equation, elasto-dynamics, and nonlinear analysis. * Designed a set of computing assignments focusing on the implementation of finite element analysis such as computing the stiffness matrix, time-integration and Newton's method.CME212, Advanced Programming for Scientists and Engineers* Developed a lecture introducing the useful features in C++11 standard (e.g. type inference, anonymous function, hash table).

* Designed and developed the website for Graduate Student Council at Stanford (http://gsc.stanford.edu) from scratch using Wordpress* Developed a web application (Wordpress Plugin) of equipment management system* Server maintenance and content management.

* Developed a new Peierls-Nabarro (PN) model via semi-discrete variational framework. Implemented a numerical solver in C++ to find the minimum energy configuration of the displacement field.* The surface effect on the mobility of a screw dislocation in the thin film has been studied by the modified SVPN model. The Peierls stress is found increased due to the reduction in elastic energy and increase in stacking fault energy by the presence of the free surface.