Muhammad Ammar Tahir
Software Engineer 💻 | Diving into Generative AI and Machine Learning Mastery!
About
An experienced software engineer who has successfully managed and directed development projects at a rapidly growing startup. Specialized in creating strong and easily expandable microservices and web applications. Skilled in using graph analytics to create innovative data visualizations and support in-depth data exploration.
Work Experience
KaleidoscopeTech Lead
Software Engineer
- ● Engineered KaleidoscopeGPT for dynamic knowledge-lake queries, delivering responsive chart outputs (PieChart, BarChart, Sankey) like EC2 distribution by Instance Type using Instruction-based fine-tuning of GPT-4
- ● Boosted the performance and scalability of the system by implementing a ReactQuery-based caching layer, slashing MongoDB queries by threefold, elevating PageSpeed score from 54 to 91, and trimming infrastructure costs by 16%
- ● Engineered a robust and scalable metric-based system utilizing Prometheus and TimescaleDB, empowering traceability and seamless handling of 10,000 metrics/sec.
- ● Accelerated graph rendering performance of Cytoscape by 10x using an optimized algorithm bringing latency down to 3 milliseconds
Education
University of Washington, Seattle
National University of Computer & Emerging Sciences
Skills
Projects
KnowledgeGraphGPT
An open-source tool with 100+ stars to convert simple text into a KnowledgeGraph using GPT-4
Essentials
An online store to sell cosmetically old but perfectly edible foods to control food wastage
Source free Domain Adaptation
Researched how transfer learning could enhance the accuracy of pre-trained BERT models
PageBook
Pagebook is a cross-platform collaborative document editor designed for team collaboration
Thulla - Card Game
A multiplayer fault-tolerant and resilient card game server
FileSystem
The Reader-Writer Lock Table is a basic implementation of lock management in a database management system.