Waleed Khan - AI/ Machine Learning Engineer

Email | LinkedIn | GitHub

Columbia University

About me

I hold a Bachelor's degree in Mechanical Engineering from Boston University and a Master's degree from Columbia University. My academic and professional journey bridges the gap between mechanical engineering and computer science, focusing on creating innovative and smart products through software.

With extensive experience in data science, embedded software engineering, and applied machine learning, I have developed a robust skill set that includes:

  • Machine Learning & Deep Learning: Proficient in developing and deploying machine learning models using frameworks such as TensorFlow, PyTorch, and Scikit-Learn.
  • Data Science & Analytics: Skilled in data preprocessing, feature engineering, and implementing data-driven solutions using Python (Pandas, NumPy), and SQL.
  • Computer Vision: Experienced with OpenCV, YOLO, and other state-of-the-art algorithms for image recognition, object detection, and segmentation tasks.
  • Natural Language Processing: Developed NLP models for text classification, sentiment analysis, and language generation using NLTK, SpaCy, and Transformer-based architectures like BERT and GPT.
  • Embedded Systems & Edge AI: Designed and optimized machine learning algorithms for deployment on resource-constrained devices using TensorFlow Lite and NVIDIA Jetson platforms.
  • Cloud & Big Data: Deployed scalable machine learning solutions on cloud platforms such as AWS, Azure, and Google Cloud, utilizing services like SageMaker, Databricks, and BigQuery.
  • Software Development: Strong programming skills in Python, C++, and Java, with experience in software engineering best practices, version control (Git), and agile methodologies.

Currently, I am an AI Engineer at Greenbacker/Helpen, where I focus on developing AI powered solutions in the renewable energy space.

Professional Journey

Explore my journey through innovative projects and impactful roles in the fields of software development and machine learning.

Generative AI/Machine Learning Consultant at Greenbacker Capital

New York, NY | Nov 2024 - Present

At Greenbacker Capital, I developed an AI-powered content management system using Python and Azure Cognitive Search, automating outbound material updates and market insights for improved diligence management, which reduced manual update time by 70%. I implemented an AI-based auto-population and data synthesis tool to streamline inbound due diligence processing, leveraging Azure OpenAI to automatically generate responses from pre-existing resources and ensure consistency across documents. I also designed AI models for intelligent task breakdown and automated data gathering, enhancing bespoke analysis assistance by structuring workflows and comparing historical analysis requests, resulting in a 70% reduction in task completion time. Additionally, I enhanced the expert content creation pipeline by developing AI-assisted draft generation tools, including automated compliance pre-checks, market integration, and brand voice consistency, accelerating content production by 40% while maintaining regulatory standards.

Software and Machine Learning Consultant at Texplora

New York, NY | Sep 2022 - Present

At Texplora, I design and implement advanced software and machine learning solutions that tackles the needs of our clients. Some projects I have worked on include live human counter for a marketing agency, prompt engineering for LLMs, automated ETL and machine learning pipelines.

Key Skills: TensorFlow, PyTorch, Scikit-Learn, Seaborn, PostgreSQL, SQL, Python, Prompt Engineering, Computer Vision, Machine Learning

Machine Learning Research Assistant at DitecT Lab, Columbia University

New York, NY | Sep 2023 - Mar 2024

As a Research Assistant at DitecT Lab, I contributed to cutting-edge research, focusing on practical machine learning solutions. I collaborated on projects involving safe trajectory planning, Diffusion models, critical scenario generation, and data retention, working closely with a team of experts. During my time there, I developed a two-stage algorithm in PyTorch to address data imbalance in regression tasks. My responsibilities also included monitoring AI models' performance and refining our approaches based on real-world data. Additionally, I co-authored a significant paper on Data Distillation, addressing critical issues in autonomous systems.

Key Skills: PyTorch, Machine Learning, Data Analysis, Python, Research Collaboration

Co-Founder and Chief Technical Officer at Frigid Dynamics

Boston, MA | Sep 2021 - Sep 2022

Initially started out as a senior design project in the mechanical engineering department at Boston University. As CTO and Co-Founder of Frigid Dynamics, I led the development of innovative robotic skis from hardware to software, focusing on software engineering and system design. I designed braking software in C++, achieving an incremental 10 RPM reduction in motor speed to enhance control. I also established rigorous system requirements and validation protocols to ensure the safety and reliability of our product. Additionally, I developed scalable software solutions and integrated feedback from stakeholders to continuously improve our designs.

Key Skills: C++, Software Engineering, System Design, Managing Partnerships, Leadership

Software Development Intern at SpaceX

Hawthorne, CA | May 2021 - Aug 2021

During my internship at SpaceX, I used my skills in Software Development and knowledge of physics to produce data based automations. The product was able to extract live sensor data and classify what course of action the engine need to take. The product also provided users with visualizations, building more confidence within the novel product. By the end of the internship, I was able to scale up the product so that it could be used across the entire company for different space vehicle programs.

Key Skills: Python, Docker, Software Automation, Version Control

AI as the future

Highlighted Projects

The following are some projects that I have worked on

Skin Disease Classification and Retrieval System

July 2024

Description: Developed a web application using Flask, allowing users to upload images for automated diagnosis. The project aimed to assist in the early detection and classification of skin diseases.

Technologies Used: Flask, EfficientNet, Data Augmentation, Image Preprocessing

My Role: Designed and implemented the web application, developed the image classification model, and created endpoints for serving disease images.

Outcome/Results: Achieved an accuracy of 90% on validation data, reduced computational time by 20%, improved model robustness and accuracy by 15%, and enhanced user experience by providing visual examples of over 500 skin diseases.

Sentiment Analysis of Airline Tweets

July 2023

Description: Administered Natural Language Processing (NLP) on 14,640 airline tweets to classify sentiments into multiple categories, aiming to understand customer sentiments and improve service.

Technologies Used: LSTM, RandomizedSearchCV, NLP Techniques

My Role: Applied NLP techniques, implemented and fine-tuned the LSTM model, and optimized hyperparameters to enhance accuracy.

Outcome/Results: Achieved an accuracy of 96%, improved model accuracy by 14% through hyperparameter tuning, and maintained a validation accuracy of around 85%.

3D Transformation, Fusion & Reconstruction of Wall-E

Feb 2023 - Mar 2023

Description: Developed a Python module for SE3 transformations and implemented high-performance functions for 3D reconstruction, aiming to enhance the efficiency and accuracy of 3D data processing.

Technologies Used: Python, Numba, GCP, TSDF Fusion, PLY File Handling

My Role: Developed and tested the Python module, implemented efficient 3D projection functions, engineered the TSDF fusion pipeline, and designed a PLY file handling class.

Outcome/Results: Improved projection speed by 10 times, achieved efficient reconstruction from 10 frames, and reduced file loading and manipulation times by 50%.

YOLO-based Vehicle Detection for Autonomous Driving

Columbia Robotics Club

Description: Developed a research paper diving into the YOLO architecture, experimented with the algorithm, and fine-tuned a YOLO model for an autonomous vehicle project to detect other vehicles on the road.

Technologies Used: YOLO, Roboflow, Data Labeling, Model Fine-tuning

My Role: Conducted in-depth research, used Roboflow to collect and label data, and fine-tuned the YOLO model for improved detection accuracy.

Outcome/Results: Enhanced the detection accuracy of the YOLO model, contributing to safer and more reliable autonomous vehicle navigation.

Get in touch

For more information, you can check out my LinkedIn and github on this page. You can also reach out!