M

MARCUS JOHNSON

About

Detail

Senior AI Engineer
Texas, United States

Contact MARCUS regarding: 
Flexible work
Starting at USD60/hour

Timeline


work
Job
school
Education

Résumé


Jobs verified_user 0% verified
  • Amazon
    Senior Ai/ML Engineer
    Amazon
    Jan 2024 - Current (2 years 4 months)
    Spearheaded the development of a Generative Al platform for automated legal document review using Amazon Bedrock (Claude V2), Lang Chain, and Pinecone, reducing manual analysis time by 60%.Built multi-format document ingestion pipelines supporting PDFs, TIFFs, and text files using LangChain Document Loader, integrated with Textract, Tesseract, and Abbyy OCR for accurate extraction from scanned inputs.Engineered custom embedding workflows with Amazon Titan Embeddings and stored vectors in Pinecone Vector Store to enable scalable semantic retrieval for Retrieval-Augmented Generation (RAG).Applied LLM MapReduce techniques for large-scale summarization tasks; fine-tuned prompts using OpenAl GPT-3.5, GPT-4, and open-source models (T5, LLaMa).Fin
  • Cognizant Technology Solutions
    Machine Learning Engineer / Ai Consultant
    Cognizant Technology Solutions
    Apr 2021 - Dec 2023 (2 years 9 months)
    Led the architecture of a Generative Al-based customer support assistant for a top U.S. bank, integrating OpenAl GPT-4 with internal CRM data using LangChain and Pinecone, improving resolution times by 40%.Developed real-time intent detection models using RoBERTa, trained on historical call center data; accuracy improved by 19% over legacy rules-based systems.Built a reusable prompt optimization framework for GPT-based tasks using dynamic context enrichment from user profiles and prior interactions.Designed and implemented an MLOps platform using Airflow + SageMaker + MLflow, managing lifecycle, deployment, and monitoring for 12+ production models.Delivered multiple client demos and technical walkthroughs, helping enterprise clients underst
  • Accenture
    Machine Learning Engineer
    Accenture
    Feb 2018 - Mar 2021 (3 years 2 months)
    Developed a legal document classification system using BERT for a global law firm; achieved F1-score of 94% across multi-label categories.Built NLP-based extraction models to automate key clause identification from contracts, integrating OCR, SpaCy, and custom regex modules.Deployed models as microservices with FastAPI, served on Azure Kubernetes Service with full monitoring and autoscaling.Implemented feature drift detection and model performance analytics using Databricks and MLflow.Supported clients in heavily regulated sectors (finance, healthcare), focusing on explainability and compliance-first model design.
  • Infosys
    Machine Learning Engineer / Data Scientist
    Infosys
    Aug 2015 - Jan 2018 (2 years 6 months)
    Developed predictive maintenance models for loT-based manufacturing clients using LSTM and Random Forest, reducing downtime by 22%.Created anomaly detection pipelines using Isolation Forest and Autoencoder architectures.Managed batch and streaming data ingestion pipelines using Apache Spark, Kafka, and Snowflake for data warehousing.Conducted cross-functional client workshops on ML strategy, model explainability, and infrastructure scaling.
Education verified_user 0% verified
  • University of Texas at Dallas
    Master of Science in Computer Science
    University of Texas at Dallas
    Aug 2012 - May 2014 (1 year 10 months)
    Richardson, TX
  • B
    Bachelor of Science in Computer Science
    Aug 2008 - May 2012 (3 years 10 months)