Tahreez

• Developed an internal RAG solution leveraging Qdrant VDB, Inngest, and LangChain, boosting employee efficiency by 50%.
• Designed and integrated an AI voice agent using LiveKit with a Proto Hologram device for a business use case.
• Implemented a cascaded pipeline, reducing inference costs by 32%, lowering perceived latency by 10%, and enhancing system reliability and scalability compared to S2S solutions.

Monash University Malaysia

• Conducted research on intelligent network behavior analysis using the CICIDS2017 dataset, achieving ~97% accuracy.
• Combined autoencoders for dimensionality reduction with reinforcement learning to model attack patterns.
• Refined autoencoder hyperparameters using grid search and Bayesian optimization, increasing detection recall to 92% while reducing false positives by 6%.

Oracle (Monash-Warwick Alliance)

• Spearheaded a 7-member team in an industry project for Oracle, developing a strategic proposal for an AI-driven sustainability platform to address the 9.2 billion tons of annual industrial waste.
• Planned AI-driven solution for waste valorization projected to boost material sorting efficiency by over 1900%.
• Formulated a business case and strategy that aligned with 6 UN SDGs and projected new revenue streams, reduced operational costs, and enhanced ESG compliance.

Misk Entrepreneurship

• Co-Founded Agentat, an AI workforce platform.

Masar Program

• Selected from thousands for intensive AI program (Transformers, LLM fine-tuning, RAG).
• Advanced to Phase 2 (top 10 teams out of 100+).
• Co-developed Wajez, AI-powered OCR solution for logistics.

CarbonZeroed

• Designed and implemented a carbon emissions analysis tool to calculate and visualize client-specific emission data.
• Developed and deployed an AI assistant using OpenAI API and Streamlit to parse and analyze internal company documents.
• Applied prompt engineering techniques to optimize LLM responses.