[yasirh002[at]gmail[dot]com]
- Phone
[408-645-9XXX]
- Location
Bay Area, CA, United States
Education
- Mode
Full-time
- Specialization
Computer Architecure
- Research
Tik Tok LLM Research
Courses:
→ Leveraging LLMs for automated Short Video Content Analysis conducted on cross-platform like YouTube and TikTok.
→ Leading TikTok data mining, prompt engineering, regression, LLM analysis, and comparing results with YouTube data.
- Mode
Full-time
Courses:
Skills
Tools
Languages
Work
- Employment type
Internship
- Work model
On-site : DALLAS, TX
Automated engineering requirements traceability for mission-critical systems and transitioned to scalable document processing solution in order to improve compliance across multiple teams.
Designed and implemented FPGA unit tests using VHDL ensuring hardware met project standards.
Validated FPGA boards leveraging real-time data collection via Com-port using oscilloscopes and breakout boxes.
Developed UI to test multiple boards to provide real-time monitoring, event logging in order to reduce testing time
- Employment type
Internship
- Work model
On-site : DALLAS, TX
Developed automation using a virtual F-16 system for increasing testing efficiency and ensuring strict performance and reliability benchmarks.
Created Python based solution to streamline onboarding of new hires and software installations reducing setup time and ensuring installation of necessary software and updates.
Common and standard Python version validation on Linux systems, ensuring version compatibility.
- Employment type
Internship
- Work model
On-site : MILPITAS, CA
Collaborated with engineers to troubleshoot and resolve PCB hardware issues, conducting failure analysis and implementing solutions to improve long-term reliability and robustness.
Optimized manufacturing processes and design validation, reducing hardware failure rates during prototypes bring-up.
Projects

Open-Source Modeling Software for Quantum Materials
- Link
Optimized the backend in Julia for high-performance computing, reducing computation time through parallelization and improving scalability for large datasets.
Developed a React-based frontend UI for user-friendly interaction, allowing remote access for team members and stakeholders.
Integrated CI/CD pipelines to automate testing and deployment, ensuring stability with each code update.
Hosted the application on a local server using ngrok, allowing remote access for team collaboration and feedback.

Stanford Pintos OS
Designed and implemented an operating system for the x86 architecture, enabling multitasking through kernel thread support.
Extended the OS with features such as user program execution, a file system, and virtual memory management, allowing it to handle more complex computational tasks.

JASH - Just Another SHell
- Link
Enhanced terminal command functionality to support dynamic background settings inspired by BASH terminal features.
Implemented argument parsing logic to enable flexible input handling for terminal commands.
Developed a method for seamlessly integrating user-defined settings into terminal operations.
Optimized command execution flow to improve overall user experience and responsiveness.

LC-3b Machine
- Link
Developed an assembler, cycle-level simulator, and a pipelined implementation of the LC-3B machine, simulating realistic computer processing.
Augmented the state machine to support advanced features like interrupt handling, exception management, and virtual memory, improving overall system efficiency.

Sign Language Identifier
Designed a real-time sign language interpreter with OpenPose Model, achieving 90% accuracy and translating them to sign language alphabet.
Scraped and curated a dataset for multi-language processing (French and Spanish) enhancing the model’s ability to recognize language-specific gestures.
Utilized advanced deep learning models, demonstrating how AI can effectively bridge communication gaps for sign language users.

Health Advocacy Agent
- Link
Built a user-friendly interface using Streamlit that simplifies complex healthcare documents, enabling users to submit feedback on policy proposals.
Processed large healthcare documents using PyPDF2 and GPT-4, reducing review time for 1,000+ page documents to under 15 minutes.
Developed an efficient AI-driven system that highlights relevant sections and explains technical terms for non-experts.
Integrated document parsing and semantic embeddings to create a streamlined review and feedback platform for healthcare advocates.

Wireless Game Controller
- Link
Developed a wireless game controller using the ARM-Cortex TM4C microcontroller, enabling real-time game control via Wi-Fi.
Implemented MQTT protocol for lightweight Wi-Fi communication, facilitating smooth data exchange between devices.
Wrote device drivers for peripheral components - ES8266 Wi-Fi module to ensure wireless connectivity of the controller.

Programmable Stopwatch/Timer
Designed a stopwatch/timer using Verilog on the Basys 3 FPGA board, implementing the RTL design methodology for hardware description.
Wrote testbenches to verify the correctness of the design, ensuring the system met timing and functional specifications.
Certificates
Explored Transformer architectures, self-supervised learning, and large language models (LLMs) like BERT and Megatron for NLP tasks such as text classification and NER. Key topics include deploying models using ONNX, TensorRT, and Triton Inference Server for optimized inference.
Focused on rapid deployment of LLM-based applications using models from Hugging Face for tasks such as semantic analysis, text generation, and conversational AI. Key concepts include state management, safe interactions, and leveraging LLM architectures for practical business solutions.
Addressed challenges in training and deploying large neural networks with tens of billions of parameters using distributed training and memory optimization techniques. Key concepts include parallelism, scaling across multiple GPUs, and deploying models with Triton Inference Server.
Focused on developing conversational AI applications with Automatic Speech Recognition (ASR) and Text-to-Speech (TTS) technologies using NeMo and Riva. Topics include integrating AI services with Kubernetes for scalable deployment and optimizing model performance in production environments.
Introduced foundational neural network concepts, including forward/backward propagation, gradient descent, and activation functions. Practical applications involved building and training deep neural networks on real-world datasets to solve complex problems.
Explored the strategic application of Six Sigma methodologies within organizations to optimize processes, reduce costs, and improve quality. Emphasized aligning Six Sigma initiatives with organizational goals and fostering a culture of continuous improvement.
Covered fundamentals of processor design, pipelining, and parallelism, as well as cache hierarchies and memory systems. Key topics include instruction set architecture, data paths, and modern advancements in multicore processors and performance optimization.