About Me
About Me
Hello, I'm Inam Ullah, a research-oriented Machine Learning Engineer and Research Fellow with over 7 years of experience specialising in safe reinforcement learning, constrained optimisation, formal verification, and learning-enabled control. Equipped with a Master of Science in Artificial Intelligence, my work bridges the gap between advanced data science architectures and rigorous safety guarantees for complex autonomous systems. I am currently seeking doctoral research opportunities to further investigate the intersection of reinforcement learning, control, optimisation, and learning-enabled autonomous systems.
Presently, I serve as a Research Fellow at Southwest Jiaotong University in Chengdu, China, where I lead research on safety-critical autonomy for railway and UAV systems. My research integrates Constrained Markov Decision Processes (CMDPs), PPO, adaptive Lagrangian optimization, and Computation Tree Logic (CTL)-based safety specifications to construct formally verified autonomy and safety-shielding mechanisms. This academic pivot builds directly upon my publication record, which includes peer-reviewed works in leading venues like IEEE Access and several high-impact manuscripts currently under review at premier journals such as IEEE Transactions on Intelligent Transportation Systems and Engineering Applications of Artificial Intelligence (EAAI).
Before fully immersing myself in safety-critical autonomous systems research, I built a robust industrial foundation across diverse data-driven domains. As a Data Scientist for a high-impact UNHCR Project, I developed machine learning-based data validation pipelines and integrated NLP systems for refugee datasets. My consulting work at TOKtAi involved deploying transformer-based sentiment models and orchestrating MLOps infrastructure using AWS cloud platforms. Additionally, my tenures at Siemens, Hyper Future, and Al-Azb Real Estate enabled me to solve complex industry challenges ranging from predictive maintenance using LSTMs and Hidden Markov Models on industrial IoT systems to processing satellite imagery via CNNs and GANs.
My comprehensive technical toolkit includes deep proficiency in Python, C++, MATLAB, and SQL, alongside frameworks like PyTorch, TensorFlow, Stable-Baselines3, and simulation tools such as ROS and Gazebo. By combining practical machine learning deployment experience with rigorous formal verification methodologies, I bring a unique, safety-first perspective to AI-driven robotics and engineering systems. I welcome connections, discussions, and collaborations with fellow researchers and institutions dedicated to pioneering reliable, learning-enabled autonomy.
Sincerely,
Inam Ullah
A Data Scientist is a highly skilled professional who combines expertise in data analysis, statistical modeling, machine learning, and formal safety verification to extract meaningful insights and solve complex, safety-critical problems. They are at the core of leveraging the power of data to drive informed decision-making, uncover patterns, and support autonomous, trustworthy system design across various high-stakes domains.
As a Data Scientist and Research Fellow specializing in Learning-Enabled Autonomy, my primary focus is to transform vast and complex datasets into actionable insights while ensuring mathematical safety guarantees for physical systems. My role is pivotal in helping organizations develop data-driven models and Safe Reinforcement Learning (Safe RL) architectures that optimize processes, enhance efficiency, and maintain absolute structural safety constraints under uncertainty.
Responsibilities of a Data Scientist
My responsibilities encompass a broad range of advanced tasks, including:
Data Cleaning and Preprocessing: Ensuring datasets are accurate, complete, and optimized for training robust deep learning and reinforcement learning agents by handling missing data, outliers, and environmental noise.
Exploratory Data Analysis (EDA): Uncovering initial patterns, correlations, and structural tendencies to guide behavioral risk modeling and predictive analysis.
Feature Engineering & Latent Representation: Designing and selecting features, alongside calibrated latent risk representations, that enhance the predictive performance and safety-awareness of machine learning models.
Building Predictive & Control Models: Developing, testing, and deploying machine learning, deep learning, and reinforcement learning frameworks for prediction, classification, anomaly detection, and constrained optimization.
Data Visualization & Verification Reporting: Creating intuitive charts, dashboards, and exact formal verification outputs to communicate predictive metrics and deterministic safety bounds effectively to stakeholders.
Expertise in Autonomous Systems & Safe RL
I have extensive experience applying advanced machine learning algorithms, deep learning techniques, and formal methods to develop reliable architectures for safety-critical physical applications:
Safe Train Movement Authority: Researching and designing constrained reinforcement learning frameworks (such as PPO integrated with adaptive Lagrangian optimization) to govern autonomous railway systems, strictly constraining unsafe actions via safety-shielding mechanisms.
UAV Safety and Risk-Aware Control: Developing anticipatory safety filtering and risk-adaptive control mechanisms for unmanned aerial vehicles (UAVs), allowing agents to navigate complex environments and execute mission-critical tasks (like autonomous landing) under severe operational constraints.
Formal Safety Verification: Compiling system specifications such as Computation Tree Logic (CTL) into formal monitors and utilizing Model Checking tools (e.g., MCMAS) to verify critical autonomous system properties exactly, guaranteeing zero specification violations even during active model exploration.
*Predictive Maintenance & Industrial IoT: Deploying predictive models (including Random Forests, LSTMs, and Hidden Markov Models) on real-time industrial sensor streams to achieve advanced anomaly and fault prediction in automated transit networks.
Technical Skills & Tools
I am proficient in leveraging popular engineering tools and frameworks to achieve these goals:
Programming & Engineering: Python, C++, MATLAB, SQL
Reinforcement Learning & Simulation: Stable-Baselines3, Ray RLlib, OpenAI Gym, ROS, Gazebo, OpenCV
Machine Learning Frameworks: PyTorch, TensorFlow, Keras, Scikit-learn, CatBoost
Data Analysis & Visualization: Pandas, NumPy, Matplotlib, Seaborn, GIS dashboards
MLOps & Cloud Platforms: Docker, MLflow, AWS, Google Cloud, Azure
With a strong foundation in electronic engineering, artificial intelligence, and formal methods, I can handle large-scale datasets effectively, build scalable cloud pipelines, and implement robust, mathematically verified machine learning models for the physical world.
Passion for Learning and Collaboration
The field of learning-enabled autonomy and data science evolves rapidly, and I am committed to staying at the forefront of advancements. I actively contribute to peer-reviewed research, publishing in leading venues and submitting cutting-edge work to premier journals like IEEE Transactions on Intelligent Transportation Systems and Engineering Applications of Artificial Intelligence (EAAI).
Collaboration is a key aspect of my role. I work closely with cross-functional research teams, industrial stakeholders, and international organizations (such as the UNHCR) to ensure that AI systems are aligned with operational objectives and absolute safety standards. My ability to translate complex statistical finding and formal verification data into clear, actionable strategies ensures that organizations can deploy learning-enabled autonomous systems confidently.
🤝 Let's Collaborate!
If you are looking for a Data Scientist and Machine Learning Engineer who can unlock the true potential of your data while engineering bulletproof safety shields for autonomous trains, vehicles, or UAVs, I invite you to explore my portfolio and reach out. Let’s work together to harness the power of trustworthy, reliable AI for your organization’s next-generation systems.