08/22/2024 updated

**** ******** ****
100 % available

Machine Learning Engineer

Cambridge, United Kingdom
Worldwide
Masters in Data Science, University of Essex
Cambridge, United Kingdom
Worldwide
Masters in Data Science, University of Essex

Profile attachments

resume.pdf

JavaScriptArtificial IntelligenceApache AirflowAmazon Web ServicesAmazon Elastic Compute CloudData AnalysisArtificial Neural NetworksCloud ComputingCluster AnalysisNvidia CUDAData CleansingInformation EngineeringExtract Transform Load (ETL)Python (Programming Language)PostgreSQLLinear RegressionMachine LearningNatural Language ProcessingNumPyOpenCVTensorFlowSoftware EngineeringData Storage TechnologiesFeature EngineeringChatbotsData SciencePyTorchFlaskLarge Language ModelsPrompt EngineeringApache SparkDeep LearningGenerative AIGoogle FirebaseBackendKerasGitLabpandasMatplotlibQuestion AnsweringScikit LearnFront End Software DevelopmentRestful ApiDocument ClassificationGPTData PipelineServerless ComputingUnsupervised LearningDatabricks
Apache Airflow, AWS EC2, AWS, Apache Spark, AI, Neural Networks, backend, back-end, chatbot, cloud infrastructure, Clustering, Data Analysis, data cleaning, data pipeline, data storage, Databricks, deep learning, LSTM, document classification , ETL, feature engineering, Firebase, Flask, front-end, ChatGPT, Generative AI, GitLab, Data Engineering, Computer Science, JavaScript, Keras , LLM, Large Language models, LLMs, Linear Regression, Machine Learning, Matplotlib, Natural Language Processing , NumPy, CUDA, OpenCV, Pandas, PostgreSQL, prompt engineering, Python, PyTorch, question answering, REST API, Scikit-Learn, Serverless, Software Development, Application Development, Tensorflow, unsupervised learning

Experience

Professional experience5 - 10 years
Freelancer since2024

Languages

EnglishFluent

Project history

Machine Learning Engineer

Bravura Solutions
Machine Learning Pipeline Development for Document Classification
* Led the design and implementation of an end-to-end machine learning pipeline for document
classification, covering all phases from data ingestion and preprocessing to model training and
fine-tuning.
* Developed the data model and ETL workflow, leveraging the Databricks platform integrated with
Apache Spark , and managed data storage using Delta Lake on AWS.
* Utilized diverse AWS clusters (compute-optimized, memory-optimized, and GPU-accelerated) for
scalable processing across multiple notebooks.
* Implemented data preprocessing and augmentation with OpenCV and created a custom annotation tool
with AWS SageMaker , ensuring data confidentiality.
* Orchestrated the workflow with Apache Airflow and employed PyTorch, Petastorm, MLFlow, and
Horovod for model training on multi-GPU distributed clusters.
* Specialized in fine-tuning and hyperparameter optimization for LLMs (BERT, LayoutLM) and
Generative AI models (Llama 2 , Amazon Titan Text), alongside prompt engineering.
* Experienced in developing LLM chatbot applications using Amazon Bedrock and Langchain, integrated
with Vector Databases .

Co-Founder | CTO

Truckkin
Co-Founder and Tech Lead at Logistics Platform Startup
* Co-founded a B2B startup offering a logistics transportation operations platform, directly managing a
development team comprising a mobile app developer and a front-end developer.
* Architected the technology stack and designed the platform's architecture, including the back-end server,
to facilitate real-time booking and tracking of vehicles for business transportation needs.
* Deployed the application back-end on AWS EC2 , ensuring seamless communication between web and
mobile applications through the REST API framework.
* Played a pivotal role in client interactions and requirements gathering, translating business needs into
technical specifications for the platform.




* Tools and Frameworks: AWS EC2 , PostgreSQL, Python, Flask REST Framework , Android (Java),
Google Maps Services , Firebase, Website Hosting.

Data Scientist

ValuStrat
Real Estate Property Valuation Analytics and Forecasting
* Conducted in-depth analytics on in-house data to deliver insights into real estate property valuations
across various regions.
* Implemented data preprocessing and trained Linear Regression and Random Forest models to accurately
predict property valuations for quarterly reporting.
* Enhanced model accuracy by applying feature engineering and cross-validation techniques,
significantly improving the RMSE.
* Tools and Frameworks: Python , Pandas, Scikit-Learn, Linear Regression, Random Forest, Flask,
Matplotlib, Seaborn.

Fashion E-commerce Application Development with AI-Powered Recommendations
* Led the development of a cross-platform mobile application using ReactNative and a web dashboard
employing basic HTML and JavaScript for a fashion e-commerce platform.
* Implemented the backend with the Flask REST framework, utilizing PostgreSQL for the database hosted
on AWS EC2, ensuring robust and scalable infrastructure.
* Integrated pre-trained VGG-16 deep learning models to analyze visual features of product images,
enhancing the recommendation engine .
* Leveraged user interactions, product specifications, and image embeddings to tailor product suggestions,
improving user experience through personalized recommendations.
* Tools and Frameworks: ReactNative, PostgreSQL , AWS EC2 , Flask REST, VGG-16, NextItNet,
CNN , Keras.

Machine Learning Engineer

Inbox Business Technologies
Enhancement of Employee Attendance System with Facial Recognition
* Tasked with upgrading an RFID-based attendance system to incorporate facial recognition, leveraging
CCTV camera feeds for real-time monitoring of employee movements.
* Established a data pipeline for capturing and storing facial images using MTCNN, ensuring efficient and
accurate face detection from video feeds.
* Employed OpenCV and Numpy for data cleaning and preprocessing, optimizing the dataset for facial
recognition.
* Applied the Chinese Whisper algorithm for clustering similar faces, facilitating accurate labeling and
identification.
* Trained a CNN model with FaceNet embeddings for facial recognition, achieving a 93% accuracy rate for
recognizing 16 employees with over 800 training images each.
* Integrated the system with CCTV via RTSP, deploying the solution on the local office server to monitor
entry and exit without manual intervention.
* Tools and Frameworks: Python , Numpy , OpenCV, FaceNet, MTCNN , YOLO v3 , Pytorch, Clustering
(Chinese Whispers Algorithm ), RTSP, CNN .

Predictive Analytics for Solar Grid Power Consumption and Battery Health
* Led a project with a Solar Solutions Provider to develop predictive models for power consumption and
battery health across their solar grid sites.
* Analyzed sensor data from various sites, conducting thorough cleaning, transformations, and
preprocessing to ensure data quality.
* Applied feature engineering to enhance data interpretability, enabling more accurate model predictions.
* Utilized RNN-based models, specifically LSTMs, to forecast power consumption and battery health,
achieving accuracy rates of over 75% and 88% respectively for selected sites.
* Addressed challenges related to missing data and site-specific calibration variations, adapting models to
improve prediction reliability.
* Tools and Frameworks: Pandas , Numpy, Matplotlib, Seaborn, LSTM (Keras), Scikit-Learn.

Founder | CEO

ParticleAI
Founder and Lead Developer of Retail Analytics Startup
* Established a startup focused on offering retail businesses detailed customer behavior reports through
in-store analytics.
* Developed an intelligent camera device powered by NVIDIA Jetson and CUDA technology, capable of
real-time monitoring of customer behavior.
* The device provided comprehensive analytics, including age, gender, emotions, movements, and
in-store time, enhancing retail strategy and customer experience.
* Implemented and deployed pre-trained deep learning models on NVIDIA Jetson to analyze video data
efficiently.
* Tools and Frameworks: Python , Numpy , OpenCV, FaceNet, MTCNN , CUDA , OpenPose , Pytorch,
Keras , CNN , Tensorflow.

Contact form

Log in to get in touch

You need to be logged in to use the contact form.

Sign upLog in