Professional Experience

Senior Manager of Data Science —  OTDS Advanced Analytics

August 2022 - Present

•  Led a multidisciplinary team of 10 data scientists and software engineers in developing and deploying scalable generative AI applications, leveraging Python, Redis, AWS, Bedrock, OpenAI, Langchain, and Haystack. Achieved significant time and cost efficiencies through innovative applications such as retrieval augmented generation and automated document generation, substantially accelerating project timelines, and streamlining processes .
• Business owner of various AI-Enabled applications that utilize Retrieval-Augmented Generation (RAG) methodologies, enabled through Redis and OpenAI to accelerate and improve information retrieval.
• Contributed to the development of novel deep learning and NLP models to translate unstructured data to useful strategic insights for the OT&DS business unit.Designed and developed a full AI-Agent framework in conjunction with OpenAI’s GPT-3.5 and GPT4.0 to augment and accelerate operational processes using the DAI framework (Decision Maker, Advisor, Investigator)
• Led the design and implementation of the organization’s DevOps framework using AWS, enabling development teams to utilize improved CI/CD pipelines, and IaC (Infrastructure as Code) to increase the reliability, frequency, and efficiency of deployments.
• Led the early investigation of the application of Large Language Models(LLMs) within the Operations Organization to identify high priority opportunities to help drive the organization’s strategy and vision of LLMs and AI-Agent Frameworks at Amgen.
• Developed AI-Agent frameworks using OpenAI, HuggingFace, and Lang-Chain to accelerate the generation of documentation (Applications, SOPs, etc…) leading to a 30% reduction in time.
• Spearheaded the development of custom prompt-engineering techniques to increase the reliability and quality of searching and information retrieval prompts up to 90%.
• Contributed to the development of various deep learning classification models using Tensorflow and Python to increase the accuracy of the regulatory classification models up to 86%. 

Data Scientist —  Operations Advanced Analytics

Feb 2020 - August 2022

• Assumed the role of AWS admin for the DIPT organization to implement best practices and ensure compliance with IT organization policies.
• Developed roadmaps, set goals and defined metrics for the group to set our strategy and vision and align with Amgen operational goals while promoting collaboration, respect and inclusion.
• Managed a team of 3 data scientist developing new forecasting and NLP applications using AWS Textract and Comprehend for the Process Development business unit.
• Led the development of novel NLP semantic searching tools (Seq2Seq, Transformers, Classification) to move our organization from data-rich to decision-smart.
• Collaborated with multifunctional partners to develop and deploy robust NLP applications using TFIDF, LSTM, and transformers (Q&A, Semantic Searching, Classification)
• Partnered with tech and business leads to develop demand forecasting models using LSTM, ARIMA, and PROPHET to increase prediction accuracies and reduce inventory   costs.
• Coached the team on managing projects, effective communication, and ensuring that  requirements were delivered within our stakeholders’ timeline.
• Established best practices to scale up NLP models within Amgen using AWS Sagemaker to increase discoverability and integrity of data in millions of healthcare documents. 

Sr Associate Data Scientist —  Digital Integration and Predictive Technologies

June 2019 - Feb 2020

• Managed a team of 2 data scientist developing new forecasting and NLP applications for the  Process Development business unit.
• Developed business intelligence monitoring platforms to detect health-related content in intellectual property filings of external competitors and advise on internal strategies.
• Designed, developed, and deployed a platform to track and report technical and business metrics to the data science senior leadership team through automated data pipelines.
• Engineered a multivariate statistical method for anomaly detection operating with a 94% accuracy (patent pending). 
• Led efforts in developing SOPs for model validation processes.
• Collaborated with software engineers to validate and take models to production. 

Associate Data Scientist —  Process Development

Nov 2017 - June 2019

• Identification of API method development conditions using machine learning.
• Developed a wide range of machine learning models to address unmet scientific and business  needs using Keras, Tensorflow, and sklearn.
• Developed a timeseries forecasting model to predict instrument LCAP operating at 88% accuracy to reduce development timelines. 
• Development of SQL-based reports for instrument utilization analysis.
• Development of ML models for instrument efficiency analysis using Python.
• Identification and quantification of API and impurities using LCMS, NMR, and GC.
• Implemented the use of advanced electronic lab notebooks strategies.

Laboratory Scientist, Minbiole Lab                                                                             

Aug 2015 - Aug 2017

• Developed a wide range of ML models and MD simulations to predict chemical properties using Python, Keras, and RDKIT.
• Developed, purified, and analyzed hundreds of Quaternary Ammonium Compounds (QACs) to create a machine learning dataset. 
• Development of MD simulation models for the prediction of QAC potency
• Application of scikit-learn machine learning models for classification using Python.
• Scale up of synthetic production of QACs in the KG scale
• Purification and characterization of compounds using LCMS and NMR
• Development and implementation of a database system for laboratory data.

Research Collaborator, High Performance Computing                                                                          

Aug 2015 - Aug 2017

• Utilized the HPC clusters to develop machine learning models for chemical potency predictions
• Explored the structural properties of pyridine-based ammonium salts using NAMD simulations
• Utilized machine learning (KMeans, KNN, Linear Regression) to identify trends in datasets
• Created Python scripts to automate all High Performance Computing (HPC) with BlueWaters