Contributed with core ideas and co-wrote awarded grants.Developing compiler analysis and optimizations using MLIR to improve high level synthesis (HLS) and HPC flows.Designed and implemented SODA-OPT for the SODA Toolchain, the first MLIR-based high-level compiler to enable system-level design of DNN models and automated flow to generate custom FPGA and ASICs designs using Bambu HLS, Vitis, and OpenROAD.

Extending the MLIR compiler framework to support efficient host/custom-accelerator driver code generation.Optimizing GPU CUDA kernels for faster 3D reconstruction of X-Ray images.Enabling end-to-end DNN accelerator simulation in TensorflowLite using SystemC.Simulating compression of memory transfers in DNN workloads.Teaching Spring 2020 class on GPU programming basics with CUDA.Profiling neural network models and researching opportunities for optimization of dense and sparse operations.Mentoring several NEU’s and National Science Foundation REU’s students during their first research experience.Implemented a communication channel between Verilog and Multi2Sim simulator’s memory model (C++) using Verilog Procedural Interface (VPI).Assisted on the design and implementation of OpenCL kernels for benchmarking purposes .Teaching assistant for Compilers Graduate Course and Embedded Robotics Undergrad Course.

Worked in a research oriented team within the Intel Labs group.Developed optimized implementations for a new hardware architecture proposed to fulfill the requirements of DARPA’s project: Software Defined Hardware (SDH).

Worked in a research oriented team within the Software and Services Group (SSG).Applied machine learning algorithms to the field of compilers, by developing new features to the LLVM framework.

Designing an electronic acquisition system for synchronized phonocardiogram (PCG) and electrocardiogram (EKG) signals. Coordinating early stages of software development for an android application interface for this system. And mentoring engineering undergraduates to assist within this project.

Structuring and coordinating software and hardware projects while managing technicians and engineers. Involved in projects of astronomic tracking systems, machinery automation, and residential/industry automation solutions

Prototyping an instrument to acquire bio-electrical signals (an Electromiograph), using a FPGA to implement the reprogrammable components control and classification algorithms on the acquired signals, the final objective is controlling an electro-mechanical prosthetic in real time.

Programming in C++ and using the OPENCV computer vision library, I have developed algorithms implementing, in Artificial Intelligence field of research, a classifier to recognizing images and the necessary performance evaluation, being the final purpose, adapting the code for a smart glass platform in order to use it in the Health-Care area.

Performing as an analog integrated circuit (IC) designer, I projected specific purpose ICs: A high voltage operational amplifier and an Electrostatic Discharge protection cell for high frequency nodes. After specifications were defined, I was in charge of implementing every project steps: Research, IC design, IC simulation, layout design and layout simulation (Using the Cadence Virtuoso tool).

Worked at Photovoltaic Inverters' research along with a PhD student that was aiming amend and suggest changes to the European Norm EN 50530:2010 - Overall Efficiency of Grid Connected Photovoltaic Inverters - so a possible adaptation could be adopted by the Brazilian National Standards Organization (ABNT) as a Regulatory Standard, with the purpose of require improved quality of same type products sold in the Brazilian market. Lab assistant task were performed, such as data acquisition in different scientific experiments and also prototyping of an low cost specter-radiometer.