🎯 AI Engineer | Researcher | Hardware & Software Enthusiast
Welcome to my GitHub!
I’m currently working as a Research Intern at MERL (Micro-Electronics Research Lab), where I've explored microprocessor design using the Chisel framework in Scala and have developed RISC-V–based processors.
Alongside this, I’m pursuing my Final Year Project (FYP) focused on mitigating hallucinations in Large Language Models (LLMs) using PyTorch.
Recently, I also started working on a startup project as an AI Expert at TheOvalLabs, where I focus on improving model performance, fine-tuning large language models, and optimizing inference pipelines for better real-world responses.
Previously, I worked as a trainee where I completed my learning journey and now I'm an AI intern.
- 💡 Passionate about Artificial Intelligence, Generative Models, and LLM optimization
- ⚙️ Experienced in both hardware design (RISC-V, Chisel) and software engineering (Java, Python, PyTorch)
- 🚀 Currently working on a GAN-based framework using LLMs to reduce hallucinations in private domain-specific models
- 🎓 Final Year Project (FYP): Reducing Hallucinations in Private LLMs using GAN-based frameworks (Python, PyTorch)
- Text-based GAN Framework using LSTMs
- Implemented a GAN model using Python and PyTorch, where LSTMs acted as the generator to generate coherent text sequences.
- Text-baed GAN Framework using LLM: GAN for Hallucination Mitigation (FYP)
- Designed a GAN pipeline integrating LLMs as Generators and DeBERTa as Discriminator to enhance factual consistency in private LLMs.
- Single-Cycle RISC-V Microprocessor
- Developed using Scala + Chisel for digital circuit design.
- 2-Stage Pipelined RISC-V Processor
- Enhanced instruction throughput and architectural efficiency.
- Gym Management System
- Developed with Java Swing (Frontend), Core Java (Backend), and MySQL (Database) for member and subscription management.
Languages:
Frameworks & Libraries:
PyTorch • Chisel • Java Swing • NumPy • Pandas • Scikit-learn
Databases:
MySQL
Others:
Git • Bash • LLM Finetuning • GAN Architectures
Building smarter, more reliable AI systems that minimize hallucinations and improve trust in private LLMs.
- 🧠 GitHub Profile
- ✉️ Email: saadmaqsood1818@gmail.com
⭐ “AI should not just think — it should reason.”