I design and deploy production-grade AI systems across computer vision, NLP, and robotics, with a focus on measurable business and research impact.
I am an experienced machine learning engineer and data scientist with a strong track record of delivering end-to-end AI/ML solutions across diverse domains.
My work spans deep learning, computer vision, natural language processing, and robotics, transforming complex requirements into reliable intelligent systems.
I bring strong software engineering practices across the full ML lifecycle—from data collection and model development to deployment, monitoring, and continuous improvement.
A focused timeline of roles where I built AI systems for research, public sector deployment, and real-time applications.
Dec 2024 – Present
Dec 2023 – Nov 2024
Dec 2021 – Nov 2023
Mar 2017 – Dec 2021
Jun 2010 – Dec 2011
Each project highlights a practical problem, technical approach, and measurable impact.
Technologies:
Description: Developed an end-to-end framework for predicting human hand motion in shared workspaces to prevent robot-human collisions. Built a custom dataset from collision scenarios, implemented 3D cell quantization and embedding models, and enabled sub-100ms predictions via efficient nearest-neighbor search.
Impact: Improved safety and throughput in collaborative robotics workflows.
Technologies:
Description: Designed 3D-PSH, a lightweight object detection approach optimized for resource-constrained platforms. Combined adaptive clustering, spatial histogram features, and BoVW + SVM classification for fast and accurate inference.
Impact: Achieved competitive detection quality with 3–5x faster inference than deep learning alternatives.
Technologies:
Description: Architected a production-grade surveillance system integrating multiple IP camera streams with face detection, recognition, tracking, and real-time monitoring dashboards.
Impact: Reduced manual check-in time by 80% and improved unauthorized-access response with real-time alerts.
Technologies:
Description: Built deep learning models for biomedical sample classification from noisy Raman spectral data, including preprocessing, augmentation, architecture design, and hyperparameter optimization.
Impact: Delivered 95%+ classification accuracy and reduced analysis time from hours to seconds.
Technologies:
Description: Built an end-to-end multilingual sentiment system (English, Urdu, Roman Urdu), including benchmark dataset creation, custom tokenization, attention-based BiLSTM, and BERT fine-tuning.
Impact: Supported real-time analysis of 100K+ tweets/day and achieved F1-score of 0.89 on Roman Urdu data.
Technologies:
Description: Developed an automated classifier for persuasive rhetoric in crowdfunding campaigns using zero-shot and few-shot prompt strategies across multiple LLMs.
Impact: Identified rhetorical features correlated with 40% higher campaign funding success.
Core tools and capabilities used to build, deploy, and scale AI systems in production and research settings.
Python (Expert), C++ (Proficient), MATLAB, SQL, JavaScript, LaTeX
TensorFlow, PyTorch, Keras, Scikit-learn, XGBoost, LightGBM, CNNs, RNNs, Transformers, GANs, AutoEncoders
OpenCV, image/video analysis, face recognition, BERT/GPT, Word2Vec, LSTM, sentiment analysis, NER
ROS, RVIZ, PCL, LiDAR processing, 3D object detection
MySQL, PostgreSQL, SQLAlchemy, Pandas, NumPy, Flask, Django, REST APIs, Docker, CI/CD
Git, Jupyter, Linux/Unix, AWS/Cloud, distributed computing
Selected publications are available via Google Scholar.
Open to collaborations in applied AI, machine learning engineering, and intelligent systems development.
Email: junaidbaber.cs@gmail.com
Google Scholar: View publications
Availability: Research collaboration, consulting, and applied AI engineering roles.