AI Researcher @ DARE-RC · LUMS
Hi, I'm
Basit Rahim
I build machine learning models, LLM tools and research software that turn messy real-world data into evidence people can act on.
Right now that means studying why children in Pakistan drop out of school, with data from more than 30,000 children. Before that, I built edge-AI backends in Rust at ShipIn, trained LLMs with RLHF at Turing and built test infrastructure at SSI.
LUMS CS · Distinction
Dean's Honor List ×3
Applied ML research
Leakage-audited models, grouped cross-validation, SHAP and proper significance tests. I care about results that hold up.
LLM engineering
RLHF training data, model evaluation, and LLM tools like an AI code reviewer that runs on every pull request.
Research software
Production FastAPI, PostgreSQL and React systems, Rust backends and Docker deployments that researchers actually use.
Experience
AI Researcher Current
DARE-RC Aug 2025 – Present- Designed, developed, and deployed an agent-based Machine Learning model to simulate factors contributing to student dropout in Pakistan, achieving 91.88% accuracy for Lahore.
- Collaborated with CITY @ LUMS to integrate GIS-based mapping and spatial analysis into the dashboard for geographic visualization.
- Developed and deployed a psychometric dashboard that can process thousands of students' data accurately.
AI Engineer
ShipIn July 2024 – Aug 2025- Conducted research and development of scalable back-end systems for ShipIn, with a focus on edge AI platforms, evaluating performance trade-offs, and ensuring robustness in real-world deployments using Rust.
- Designed and experimented with real-time data pipelines and novel database optimization strategies to enable efficient large-scale data processing, storage, and retrieval for research-driven analytics.
Python & Data Engineer
Turing Aug 2024 – May 2025- Researched training of large language models for coding and data science using Reinforcement Learning with Human Feedback (RLHF), with emphasis on model alignment, robustness, and reliability.
- Implemented and evaluated systematic methodologies for assessing model-generated solutions to software engineering problems, analyzing failure cases and proposing research-driven improvements.
Software Engineer
Strategic Systems International Jul 2023 – Aug 2024- Developed a research-grade test infrastructure for ASMPT's line control systems, enabling automated validation, large-scale simulations, and reproducible experimentation.
- Applied simulation frameworks with RabbitMQ for real-time communication studies and designed REST APIs to support extensible integration and remote experimentation on line control units.
Projects
Research & engineering case studies
More projects
Equity Equation Project
An interactive data-driven application exploring equity in education, built with Streamlit. Analyzes and visualizes educational data to support evidence-based policy decisions.
Plutus
A full-stack financial application with a deployed web interface. Built with modern web technologies for seamless user experience.
LLM-Based SOAP Note Generator & Evaluator
An LLM-powered application that automatically generates and evaluates SOAP (Subjective, Objective, Assessment, Plan) clinical notes, deployed on Vercel.
AI Tutor
Developing an AI-powered tutor for a large school chain in Pakistan, providing personalized learning based on the school's curriculum. Leading the pilot with plans to scale across 150+ schools.
Early Detection of Third-Party Abuse on Twitter
Developed a machine learning model to detect abusive third-party applications on Twitter by analyzing early activity. Conducted a longitudinal study contributing to automated abuse detection systems.
Education
Lahore University of Management Sciences (LUMS)
Bachelors of Science in Computer Science
Aug 2019 – June 2023