Abdul Mannan

Contributing to an AI-driven Sales & Marketing Intelligence platform used by Fortune 500 companies, focusing on scalable backend services, GenAI workflows, and production-grade systems combining LLMs, vector search, and retrieval pipelines. Collaborating with cross-functional teams to architect and optimize AI-driven backend services.

• Applying predictive modeling and data-driven diagnostics to improve reliability of solid-state transformers, reducing
downtime and operational risks.
• Building real-time anomaly detection system, showcasing ability to monitor performance signals and optimize efficiency.
• Designing insight dashboards from telemetry data, supporting leadership in decision-making and operational strategy.

• Built game apps in Python (Kivy) for MyndBoard's digital activity table, used by elderly and cognitively impaired residents in care facilities
• Followed a design-first workflow, delivering the project in structured milestones with clean backend architecture and responsive, touch-optimized UI
• Added multi-lingual support and fixed bugs through iterative testing to improve stability and usabilit

• Built a SaaS LMS using React.js and Supabase with role-based access (admin, user, student, member)
• Used Brevo for sign-up/sign-in and marketing emails; integrated Stripe for payments, subscriptions, and access control
• Built an admin panel to toggle courses on/off, set discounted pricing, and manage videos, transcripts, and course docs.
• Deployed on Vercel.

• Built an interactive 3D web-based product configurator for industrial screw compressors using Vue.js and Three.js, enabling clients to visually explore CAD models directly in the browser without specialized CAD software

• Engineered a real-time material customization system that lets users dynamically recolor individual compressor components (piping, engine/compressor housing) and toggle part

• Gained hands-on experience in NLP, pre-processing, text-to-vector embeddings, and BoW, improving text classification accuracy by 15-20%. Learned CNNs, RNNs, LSTMs, and GRUs, enhancing model performance and reducing training time by 10-15% through optimized architectures and hyperparameter tuning.
• Specialized in transformer architectures like BERT, GPT, and encoder-decoder, optimizing performa

I've had the privilege of working on diverse projects, ranging from custom desktop applications to well-built web apps.

As a seasoned freelancer, I've delivered high-quality solutions on Fiverr, earning recognition as a Level 1 Seller and Fiverr's Choice.

Pursued a Bachelor of Science in Computer Science, covering coursework in algorithms, machine learning, and artificial intelligence.