Adil is available for hire

Adil Mubeen

Verified Expert in Engineering

Machine Learning Engineer and Developer

San Francisco, CA, United States

Toptal member since December 18, 2024

Expertise

Computer Vision Machine Learning Deep Learning Artificial Intelligence NLP RAG Full-stack OpenAI Web Development LLM PyTorch TensorFlow Python DevOps Kubernetes

Bio

Adil is an experienced machine learning engineer who specializes in AI, deep learning, large language models (LLMs), Python, C/C++, retrieval-augmented generation (RAG), computer vision, and natural language processing (NLP). He's proficient in developing and deploying scalable ML systems using TensorFlow and PyTorch. Adil consistently delivers innovative, intelligent solutions tailored to clients' needs.

Portfolio

Digiverse

ChatGPT, Machine Learning, Natural Language Processing (NLP), Computer Vision...

Aridian Technologies

Artificial Intelligence (AI), Computer Vision...

Malta Themes

Artificial Intelligence (AI), Machine Learning, Deep Learning, Computer Vision...

Experience

Retrieval-augmented Generation (RAG) - 7 years
Machine Learning - 7 years
Python - 7 years
Generative Artificial Intelligence (GenAI) - 7 years
Artificial Intelligence (AI) - 7 years
Large Language Models (LLMs) - 5 years
ChatGPT - 3 years
LlamaIndex - 3 years

Availability

Full-time

Preferred Environment

PyTorch, Large Language Models (LLMs), Computer Vision, Machine Learning, Deep Learning, SQL, Python, Artificial Intelligence (AI), C, DeepStream SDK

The most amazing...

...AI customer support platform I've built leverages LLMs and NLP to automate responses, enhance accuracy, and improve user satisfaction.

Work Experience

Machine Learning Engineer

2022 - 2024

Digiverse

Led a team of six ML engineers and worked on the production-ready deployment of a construction CV system, including diffusion models for real-time rendering of safety compliance and unsafe behaviors that reduced accident rates by 12%.
Worked hands-on on diffusion models, especially model inference optimization for retail video analytics and Azure deployments, using behavior analysis to monitor customer engagement and optimize store layouts, which increased profit by 10%.
Developed and optimized LLM algorithms and infrastructure for an AI customer support platform, achieving an 80% reduction in response time and a 65% decrease in operational costs through parallel computing, CI/CD pipelines, and GPU acceleration.

Technologies: ChatGPT, Machine Learning, Natural Language Processing (NLP), Computer Vision, Python, Amazon Web Services (AWS), Data Pipelines, AWS Data Pipeline Service, ChatGPT Prompts, ChatGPT API, Large Language Models (LLMs), Multiagent Generative Systems (MAGs), Deep Learning, Deep Reinforcement Learning, Deepfake, Big Data, Databases, Data Science, Data Scraping, Data Scientist, Agile Data Science, Generative Artificial Intelligence (GenAI), You Only Look Once (YOLO), YOLOv5, Large Language Model Operations (LLMOps), Machine Learning Operations (MLOps), DevOps, C++, C, C#, iOS, Swift, TensorFlow, PyTorch, PyTorch3D, PyTorch Lightning, Python 3, Git, GitLab, CI/CD Pipelines, GitLab CI/CD, Retrieval-augmented Generation (RAG), Data Classification, AI Data Classification, Docker, Kubernetes, Kubernetes Operators, Azure, Azure SQL, SQL, MySQL, NoSQL, PostgreSQL, MongoDB, Google Cloud, BigQuery, Google BigQuery, Google Bigtable, Amazon Marketplace Web Service (MWS), Amazon SageMaker, Chatbots, Speech to Text, Text to Speech (TTS), Speech to Text AI, Image to Text, Text to Image, NumPy, Pandas, Transformers, Data Transfers, Data Cleaning, Data Cleansing, Data Annotation, Data Analysis, Data Analytics, Oracle Database, Models, Modeling, Tableau, Spark, SpaCy, Spark ML, Spark NLP, Spark SQL, Data Lakes, Data Labeling, Web Development, APIs, Back-end, Back-end Development, DeepStream SDK, Data Inference, Azure Machine Learning, Supervised Machine Learning, Amazon Machine Learning, Object Detection, Microsoft Excel, Business Analysis, Business Analytics, IT Business Analysis, AI Agents, AgentGPT, Django, Flask, Matplotlib, Scikit-learn, FastAPI, Infrastructure, Data Interfaces, Bayesian Inference & Modeling, Deepstream.io, Localization, LoRa, Tesseract, Stable Diffusion, Snowflake, Bioinformatics, Model Tuning, DeepSeek, Llama, Model Context Protocol (MCP), Prompt Engineering, Agentic AI, AutoGen, Autonomous AI, BERT, Contract Lifecycle Management (CLM), Falcon, Hugging Face, Hugging Face Transformers, Legal Technology (Legaltech), Long Short-term Memory (LSTM), Natural Language Understanding (NLU), PEFT, Qwen, PLAIF, Semantic Kernel (SK), Synthetic Data Generation, Quantitative Analysis, Statistical Analysis, Banking & Finance, Finance, Fintech, Healthcare IT, Healthcare, CTO, Fine Tunning, Pattern Recognition, Vector Databases, Fine-tuning, Medical Imaging, Sentiment Analysis, OpenCV, Image Recognition, PDF, Algorithms, Data Architecture, Data Security, Blockchain, Encryption, Gmail API, Slack API, OpenAI API, Node.js, AI Tools, React, Data Engineering, Amazon EKS, Amazon Elastic Container Service (ECS), Amazon S3 (AWS S3), AWS IoT, AWS Lambda, Database Optimization, PySpark, Chatbot Conversation Design, Conversational AI, AI Prompts, Dialogflow, Whisper, Natural Language Generation (NLG), Knowledge Bases, Neural Networks, Video Analytics, Human Activity Recognition (HAR), Gemini, Low Latency, Audio Streaming, Performance, Pinecone, Statistical Methods, Statistical Data Analysis, Pricing, Pricing Models, Google Sheets API, Next.js, REST APIs, Tailwind CSS, Dashboards, AI Compliance Agents, AI Automation, Time Series Analysis, Trading, Financial Modeling, Jupyter, Llama 3, Agile Software Development, Software Architecture, LangGraph, AI Integration, Drones, Robotics, Startups, ChromaDB, Redis, Zoom, Graph Databases, Cypher, Neo4j, Data Management, Lead Generation, Presentations, Recommendation Systems, Sales

Machine Learning Engineer

2019 - 2022

Aridian Technologies

Built ML models and data pipelines to ensure workers' safety on construction sites, leading to a 32% reduction in workplace injuries.
Contributed to object detection, human pose estimation, and activity recognition.
Developed a resume management system that streamlined the candidate screening process, reducing analysis time by 90%. Also implemented NLP algorithms to improve resume parsing accuracy, increasing relevant candidate matches by 25%.
Created a conversational AI solution that helped over 3,000 children learn Spanish, English, and Chinese.

Technologies: Artificial Intelligence (AI), Computer Vision, Natural Language Processing (NLP), Machine Learning, Python, Python 3, Amazon Web Services (AWS), Deep Learning, Data Pipelines, ChatGPT, ChatGPT API, Chatbots, AI Chatbots, Large Language Models (LLMs), Large Language Model Operations (LLMOps), Multiagent Generative Systems (MAGs), PyTorch, TensorFlow, Generative Artificial Intelligence (GenAI), Multimodal GenAI, You Only Look Once (YOLO), Machine Learning Operations (MLOps), DevOps, TensorBoard, Retrieval-augmented Generation (RAG), Data, Big Data, Databases, Data Cleaning, Data Cleansing, Docker, Kubernetes, AWS Data Pipeline Service, ChatGPT Prompts, Deepfake, Deep Reinforcement Learning, Reinforcement Learning, Reinforcement Learning from Human Feedback (RLHF), BigQuery, Google BigQuery, Google Bigtable, BigTable, Data Science, Data Scraping, Data Scientist, Agile Data Science, Data Lakes, Data Labeling, C++, C, C#, iOS, Web, Web Development, iOS App Development, PyTorch3D, PyTorch Lightning, Git, GitLab, CI/CD Pipelines, GitLab CI/CD, Data Classification, Azure, Azure SQL, SQL, MySQL, NoSQL, PostgreSQL, MongoDB, Oracle, SQL Server 7, Image to Text, Text to Image, NumPy, Pandas, Matplotlib, Scikit-learn, Object Detection, Text Classification, Classification, Image Classification, Microsoft Excel, AgentGPT, AI Agents, Django, Flask, FastAPI, Infrastructure, Bayesian Inference & Modeling, Bayesian Statistics, Localization, LoRa, Tesseract, ComfyUI, Blender, Firebase, Snowflake, Bioinformatics, Agentic AI, AutoGen, Autonomous AI, BERT, Contract Lifecycle Management (CLM), Falcon, Hugging Face, Hugging Face Transformers, Legal Technology (Legaltech), Long Short-term Memory (LSTM), Natural Language Understanding (NLU), PEFT, Qwen, PLAIF, Semantic Kernel (SK), Synthetic Data Generation, Quantitative Analysis, Statistical Analysis, Agile, FIG, Healthcare, CTO, Pattern Recognition, Predictive Modeling, Vector Databases, Fine-tuning, Medical Imaging, Sentiment Analysis, OpenCV, Image Recognition, Algorithms, PDF, Data Archiving, Data Architecture, Blockchain, Data Security, Encryption, Gmail API, Slack API, OpenAI API, Node.js, React, AI Tools, Data Engineering, Amazon EKS, Amazon Elastic Container Service (ECS), AWS IoT, AWS Lambda, Amazon S3 (AWS S3), Database Optimization, Dialogflow, Whisper, Chatbot Conversation Design, AI Prompts, Conversational AI, Natural Language Generation (NLG), Knowledge Bases, Neural Networks, Video Analytics, Human Activity Recognition (HAR), Gemini, AI enabled Applications, Low Latency, Audio Streaming, Statistical Methods, Statistical Data Analysis, Pinecone, Performance, Pricing, Pricing Models, AI Automation, Next.js, Google Sheets API, REST APIs, Tailwind CSS, Dashboards, Time Series Analysis, Trading, Financial Modeling, Jupyter, Llama 3, Agile Software Development, Software Architecture, LangGraph, AI Integration, Drones, Robotics, Startups, ChromaDB, Redis, Zoom, Neo4j, Cypher, Graph Databases, Lead Generation, Data Management, Presentations, Recommendation Systems, Sales

Machine Learning Engineer

2018 - 2019

Malta Themes

Implemented a real-time fraud detection system with deep learning models in PyTorch, achieving 98% accuracy and preventing significant financial losses.
Created a computer vision-based quality control system utilizing You Only Look Once (YOLO) and TensorFlow, improving defect detection by 30% and reducing manual inspection costs.
Contributed to memory, communication, and data loading optimization for a ticket-selling recommendation system that increased sales by 16%.

Technologies: Artificial Intelligence (AI), Machine Learning, Deep Learning, Computer Vision, Natural Language Processing (NLP), You Only Look Once (YOLO), TensorFlow, Python, Python 3, ChatGPT, ChatGPT API, ChatGPT Prompts, Azure Machine Learning, Amazon Machine Learning, Amazon Web Services (AWS), Data Pipelines, Large Language Models (LLMs), Multiagent Generative Systems (MAGs), AgentGPT, AI Agents, PyTorch, PyTorch Lightning, PyTorch3D, Artificial Neural Networks (ANN), Generative Artificial Intelligence (GenAI), Multimodal GenAI, Multimodal Models, Big Data, Databases, Data Science, Data Analysis, Data Analytics, SQL, NoSQL, PostgreSQL, Oracle, MongoDB, Optical Character Recognition (OCR), Large Language Model Operations (LLMOps), DevOps, Retrieval-augmented Generation (RAG), LangChain, Data Classification, Docker, Kubernetes, Azure, Azure SQL, Google Cloud, Amazon SageMaker, ML Pipelines, AI Pipeline, GitLab, GitHub, Chatbots, OpenAI, OpenAI API, OpenAI Gym, NumPy, Pandas, C++, C, C#, Speech to Text, Text to Speech (TTS), Speech to Text AI, Speech Recognition, Speech Synthesis, Machine Translation, Reinforcement Learning, Deep Reinforcement Learning, Localization, LoRa, Snowflake, Bioinformatics, Quantitative Analysis, Statistical Analysis, Healthcare, CTO, Predictive Modeling, Pattern Recognition, Vector Databases, Fine-tuning, Medical Imaging, Image Recognition, OpenCV, Sentiment Analysis, Algorithms, PDF, Data Architecture, Data Security, Encryption, Blockchain, Gmail API, Azure OpenAI Service, Slack API, Amazon EKS, Amazon Elastic Container Service (ECS), AWS IoT, AWS Lambda, Amazon S3 (AWS S3), Database Optimization, Conversational AI, AI Prompts, Chatbot Conversation Design, PySpark, Dialogflow, Whisper, Natural Language Generation (NLG), Knowledge Bases, Neural Networks, Video Analytics, Human Activity Recognition (HAR), Gemini, Low Latency, Audio Streaming, Pinecone, Statistical Methods, Statistical Data Analysis, Performance, Pricing, Pricing Models, Tailwind CSS, Dashboards, AI Automation, Next.js, REST APIs, Google Sheets API, ai compliance agents, Time Series Analysis, Trading, Financial Modeling, Jupyter, Llama 3, Agile Software Development, Software Architecture, ChromaDB, Redis, Zoom, Graph Databases, Neo4j, Cypher, Sales, Lead Generation, Lead Management, Presentations, Recommendation Systems

Experience

Anti-theft Security Application

A Python, C, and NVIDIA DeepStream Anti-theft security project in which I implemented a real-time camera inference pipeline. The pipeline processed live video streams, integrating object detection and action recognition models, optimized with TensorRT for GPU acceleration to achieve low latency and high throughput. Python was used to manage the pipeline and integrate deep learning models, while performance-critical components, such as custom plugins for DeepStream, were written in C to optimize memory management and computational efficiency.

The system leveraged CUDA for parallel processing, enabling efficient handling of high-resolution video streams on edge devices like NVIDIA Jetson. Preprocessing tasks were designed to minimize data pipeline bottlenecks, including resizing, normalization, and batching. Real-time alerts and metadata logging were incorporated to detect and flag suspicious activities, ensuring timely responses and detailed incident tracking for further analysis.

AI Customer Support Agent

One of my most notable achievements is developing an advanced AI customer support platform engineered entirely in Python. This system integrates Twilio to handle incoming calls, seamlessly converting speech to text using state-of-the-art speech recognition technologies. The transcribed text is then processed by ChatGPT and enhanced with RAG via LangChain, ensuring accurate and contextually relevant responses. The platform then delivers these responses to the user, employing Eleven Labs' text-to-speech capabilities and enabling natural and fluent voice interactions.

The entire solution is containerized using Docker and deployed on Google Cloud Platform (GCP) with Kubernetes, ensuring scalability, reliability, and efficient resource management. This end-to-end system automates customer support interactions, significantly enhancing response accuracy and user satisfaction. By leveraging large language models LLMs and NLP, the platform provides a scalable and intelligent support system tailored to meet diverse client needs.

AI-powered Health Care Assistant

I spearheaded the creation of an AI-powered virtual healthcare assistant that leverages NLP and computer vision to interpret patient inquiries and analyze medical images. This platform processes voice and text inputs to provide accurate medical information and diagnostic support, enhancing patient care and streamlining healthcare services.

Deployed on Azure with Kubernetes orchestration, the system ensures high availability and scalability, while Docker containers facilitate seamless updates and maintenance. By integrating LLMs with specialized medical datasets, the assistant delivers reliable and context-aware responses, significantly improving patient engagement and operational efficiency for healthcare providers.

AI-powered Construction Worker Safety System

I developed and optimized ML models and data pipelines to enhance worker safety on construction sites, reducing workplace injuries by 32%. The system monitors site environments in real time by utilizing You Only Look Once (YOLO) for precise object recognition and advanced activity recognition algorithms. By integrating computer vision techniques with NLP, the platform can detect unsafe behaviors and potential hazards, ensuring immediate compliance with safety standards.

The solution, containerized with Docker for seamless deployment, leverages deep learning frameworks such as TensorFlow and PyTorch. Deployed on Google Cloud Platform (GCP) using Kubernetes, the infrastructure ensures scalability, reliability, and efficient resource management. The implementation includes CI/CD pipelines and GPU acceleration to optimize performance and facilitate rapid updates. This comprehensive approach automates safety compliance monitoring and provides actionable insights through robust data pipelines, significantly enhancing overall site safety standards.

Learn New Language Chatbot for Students

https://meglanguages.com/

An innovative language learning platform for children, enabling them to learn English, Spanish, and Chinese interactively. Powered by ChatGPT’s LLM, the platform uses cutting-edge AI to provide personalized language instruction. Built with Python for back-end logic, Node.js for API handling, and Angular for a seamless web interface, it offers an engaging user experience. Features include real-time text-to-speech and speech-to-text capabilities, enhancing pronunciation and conversational practice. Deployed on Azure using Kubernetes and Docker, the platform ensures scalability and reliability for its 3,000+ active student base. Designed for young learners, it combines interactive lessons with adaptive AI, creating a fun and effective way to master new languages.

AI Phone Assistant

https://aozic.ai/

Aozic is a B2B AI solution that automates customer support calls, booking, order taking, and more, acting as a virtual agent. It saves businesses time, improves customer satisfaction, and integrates seamlessly, offering a 24/7 solution to boost productivity and growth.

Deep Reinforcement Learning Application for Port Automation

In a previous large-scale industrial project, I was deeply involved in developing and deploying a multi-agent Deep Reinforcement Learning (DRL) application for port automation. This endeavor focused on optimizing complex terminal operating systems where thousands of containers, dozens of autonomous cranes, trucks, and scheduling agents needed to coordinate in real-time. The core challenge involved managing continuous state spaces, ensuring safety-critical constraints, and optimizing container movements, berth allocation, and yard operations. My contributions included customizing actor-critic methods, devising novel reward shaping techniques to address sparse reward environments, and implementing hierarchical RL approaches to effectively manage the multi-timescale nature of port operations, ranging from immediate crane movements to long-term vessel scheduling. The real-time data pipelines processed live vessel tracking data, dynamic weather conditions, and real-time equipment status. The infrastructure was designed to serve RL policy decisions with sub-second latency requirements, leveraging technologies like Apache Kafka for high-throughput data streaming.

Education

2022 - 2024

Master's Degree in Artificial Intelligence

Monroe University - New York, USA

Certifications

JUNE 2019 - PRESENT

Deep Learning Specialization

Coursera

Skills

Libraries/APIs

PyTorch, TensorFlow, NumPy, Pandas, PyTorch Lightning, PyTorch3D, OpenAI API, Spark ML, Spark NLP, Amazon Marketplace Web Service (MWS), SpaCy, Matplotlib, Scikit-learn, Camera API, Hugging Face Transformers, OpenCV, Gmail API, Slack API, Node.js, React, PySpark, Google Sheets API, REST APIs

Tools

ChatGPT, You Only Look Once (YOLO), Tableau, Microsoft Excel, GitLab CI/CD, GitLab, Git, Spark SQL, OpenAI o1, Azure Machine Learning, Kubernetes Operators, BigQuery, Amazon SageMaker, TensorBoard, GitHub, OpenAI Gym, Blender, ComfyUI, DeepSeek, AWS CLI, Azure OpenAI Service, Amazon EKS, Amazon Elastic Container Service (ECS), AI Prompts, Dialogflow, Whisper, Claude, Jupyter, Zoom

Languages

Python, C++, SQL, C, JavaScript, TypeScript, C#, Swift, Python 3, PHP, HTML, CSS, Snowflake, Falcon, Cypher

Frameworks

LlamaIndex, Spark, Hadoop, Django, Flask, Deepstream.io, AutoGen, Next.js, Tailwind CSS, LangGraph

Paradigms

DevOps, Agile, Model Context Protocol (MCP), Synthetic Data Generation, Automation, Agile Software Development

Platforms

Azure, Kubernetes, Amazon Web Services (AWS), Docker, Polyaxon, Oracle Database, Web, iOS, Oracle, WordPress, Amazon, Firebase, Blockchain, AWS IoT, AWS Lambda

Storage

Data Pipelines, Google Cloud, NoSQL, MongoDB, MySQL, Databases, AWS Data Pipeline Service, Azure SQL, PostgreSQL, Google Bigtable, Data Lakes, BigTable, SQL Server 7, Amazon S3 (AWS S3), Redis, Graph Databases, Neo4j

Industry Expertise

Bioinformatics, Banking & Finance, Healthcare

Other

Large Language Models (LLMs), Computer Vision, Machine Learning, Deep Learning, Generative Artificial Intelligence (GenAI), Speech to Text, Artificial Intelligence (AI), Data Science, Natural Language Processing (NLP), Machine Learning Operations (MLOps), Retrieval-augmented Generation (RAG), Full-stack, OpenAI, Web Development, DeepStream SDK, ChatGPT Prompts, Multiagent Generative Systems (MAGs), ChatGPT API, Data Classification, CI/CD Pipelines, Chatbots, Transformers, Models, APIs, Large Language Model Operations (LLMOps), Back-end Development, Data Inference, Machine Learning Algorithms, Object Detection, Business Analysis, Business Analytics, Data Analysis, Data Analytics, Text to Speech (TTS), API Integration, Reinforcement Learning, Deep Reinforcement Learning, Data Scientist, Semantic Analysis, Architecture, Leadership, AI Agents, AgentGPT, Big Data, Deep Neural Networks (DNNs), Data Scraping, Data, Optical Character Recognition (OCR), Modeling, Virtual Machines, Speech to Text AI, Azure Text to Speech, IBM Watson Speech to Text, Agile Data Science, AI Research, Deepfake, YOLOv5, AI Data Classification, Google BigQuery, Image to Text, Text to Image, Data Transfers, Data Cleaning, Data Cleansing, Data Annotation, Data Labeling, Back-end, Supervised Machine Learning, Amazon Machine Learning, IT Business Analysis, AI Chatbots, FastAPI, Multimodal GenAI, Reinforcement Learning from Human Feedback (RLHF), iOS App Development, Text Classification, Classification, Image Classification, Full-stack Development, Speech Recognition, Machine Translation, Neural Machine Translation, Artificial Neural Networks (ANN), Multimodal Models, LangChain, ML Pipelines, AI Pipeline, Speech Synthesis, AWS Jam, ElevenLabs Solutions, Generative Pre-training Transformer, Azure IoT, Website Data Scraping, Feature Engineering, Infrastructure, Data Interfaces, Bayesian Inference & Modeling, Bayesian Statistics, Cloud, GPU Computing, NVIDIA A100 Tensor Core GPU, Serverless GPUs, Generative Pre-trained Transformers (GPT), Localization, LoRa, Tesseract, Stable Diffusion, Model Tuning, Llama, Prompt Engineering, Agentic AI, Autonomous AI, BERT, Contract Lifecycle Management (CLM), Hugging Face, Legal Technology (Legaltech), Long Short-term Memory (LSTM), Natural Language Understanding (NLU), PEFT, Qwen, PLAIF, Semantic Kernel (SK), Quantitative Analysis, Statistical Analysis, Finance, Fintech, FIG, Healthcare IT, CTO, Fine Tunning, Pattern Recognition, Predictive Modeling, Vector Databases, Fine-tuning, Medical Imaging, Sentiment Analysis, Image Recognition, PDF, Algorithms, Data Architecture, Data Security, Encryption, Data Archiving, AI Tools, Data Engineering, Database Optimization, Chatbot Conversation Design, Conversational AI, Natural Language Generation (NLG), Knowledge Bases, Neural Networks, Video Analytics, Human Activity Recognition (HAR), Gemini, AI enabled Applications, Low Latency, Audio Streaming, Performance, Pinecone, Statistical Methods, Statistical Data Analysis, Pricing, Pricing Models, Dashboards, AI Compliance Agents, AI Automation, ai compliance agents, Time Series Analysis, Trading, Financial Modeling, Llama 3, Software Architecture, AI Integration, Drones, Robotics, Startups, ChromaDB, Data Management, Lead Generation, Presentations, Recommendation Systems, Sales, Lead Management

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring