REHMAN MALLAH

• Created RAG system to intake more than 500-page documents to analyse with LLMs overcoming their context length constraints
• Built inventory management system including 3-month sales averages to automate product-specific reorder decisions
• Developed web scraping tool to automate product data extraction of more than 10,000 items from major supermarket websites, with scheduled 4-hour updates ensuring real-time database accuracy
• Built multiple specific-purpose parsers and analysers for PDFs with custom extraction techniques and OCR integration
• Administered AWS IAM policies and EC2 servers, troubleshooting system faults and enforcing policy compliance

• Elevated software quality through automated integration testing by developing Python and Bash programs for rigorous testing procedures; reducing manual testing by 10 hrs/week
• Enhanced software delivery processes by developing CI/CD pipelines in Azure DevOps with continuous inspection tools
• Transformed cloud architecture efficiency by orchestrating Kubernetes clusters alongside Docker containers, significantly lowering typical application rollout durations on AWS and Azure
• Performed cloud penetration tests detecting critical server-side vulnerabilities affecting user dashboards and databases
• Configured and managed dozens of testing and production on-site virtual machines with VMware ESXi ensuring operational stability and availability
• Diagnosed and resolved technical issues for more than 50 clients via Webex calls, ranging from basic installation to LDAP and AD issues

• Researched over a dozen AI algorithms and machine learning techniques to formulate clang-based compiler-optimisation challenges into an AI-driven solution
• Designed and programmed a metaheuristic approach using simulated annealing to optimise compiler flags, achieving a 20% improvement in program execution speed
• Authored a 60-page technical research paper demonstrating results on par with industry-leading studies in the field

• Led a 6-member team to deliver a portfolio management product for the client, Telefonica's Wayra, within 2 months
• Adopted a hybrid Waterfall-Agile development model and engaged in continuous communication with the client over two months, organising meetings to gather requirements and addressing emerging issues
• Integrated AirTable API for real time syncing between client databases of 1000+ records and the web application
• Engineered full database functionality: CRUD operations (search/filter/add/delete) with robust error handling

• Engineered an AI-driven job search agent using Selenium and LLM integration to autonomously navigate LinkedIn, process user CVs via PDF upload, and prioritise relevant opportunities—eliminating manual search workflows
• Created smart CV parser for structured data extraction, auto-populating class variables to generate personalised profiles
• Automated navigation for job post listing and authentication — achieving end-to-end autonomy with zero intervention

• Implemented the complete decoder transformer with sequential Neural Networks to create GPT LLM architecture
• Tokenised input text using cl100k-base, converted tokens into vector embeddings, and applied a multi-head attention mechanism to generate decoder inputs for advanced sequence processing
• Trained the neural network using a massive 40GB open-source replication of the WebText dataset from OpenAI
• Optimised model performance through hyperparameter tuning and gradient clipping methods