Horia Mărgărit, Developer in San Jose, CA, United States
Horia is available for hire
Hire Horia

Horia Mărgărit

Verified Expert  in Engineering

AI Engineer and Consultant Developer

San Jose, CA, United States

Toptal member since November 16, 2021

Bio

Horia's industry experience extends across digital health, consumer finance, internet search, engineering software, and transportation logistics. He has over 10 years of expertise applying AI methods and approaches to tackle businesses problems. Horia's predictions for business applications of AI have been featured in CIO magazine and Forbes. He earned a master of science degree in statistics at Stanford University and two BAs specializing in cognitive and computer science at UC Berkeley.

Portfolio

Tropicana Brands Group
JAX, Transfer Learning, LoRa, AI Agents, Light LLMs...
US Naval Nuclear Lab & Stanford Wehab Lab
Quantum Computing, Quantum Circuits, Quantum Machine Learning...
PepsiCo
Data Science, Leadership, Machine Learning, Software Architecture...

Experience

  • Deep Learning - 11 years
  • Explainable Artificial Intelligence (XAI) - 7 years
  • Cognitive Computing - 5 years
  • Cognitive Science - 5 years
  • Psychology & Mental Health - 3 years
  • Cognitive Behavioral Therapy (CBT) - 3 years
  • Multi-task Learning - 2 years
  • Meta-learning - 2 years

Availability

Part-time

Preferred Environment

Linux, Jupyter Notebook

The most amazing...

...thing I've developed is an AI system for personalized behavioral health treatment. It was published in the IEEE journal. It won an honorable mention at ACM CHI.

Work Experience

Team Lead | Machine Learning

2024 - 2024
Tropicana Brands Group
  • Built time-series forecasting models for sales, supply chain, and integrated business planning (IBP), benchmarking GRU, Transformer, and long short-term memory (LSTM) models within Apache Spark (Databricks).
  • Developed an AI assistant using reinforcement learning from human feedback (RLHF) and benchmarked it against PEFT using LoRA for optimized response quality.
  • Led algorithm development for marketing and trade promotion optimization. Developed contextual multi-armed bandits (MAB) using the upper confidence bound (UCB).
  • Benchmarked deep learning models and gradient-boosted trees for context learning.
Technologies: JAX, Transfer Learning, LoRa, AI Agents, Light LLMs, Large Language Models (LLMs), Open-source LLMs, Contextual Bandits, Multi-armed Bandits (MABs), Non-generative AI, Transformer Models, Long Short-term Memory (LSTM), LSTM Networks, LSTM, Gated Recurrent Unit (GRU), Recurrent Neural Networks (RNNs), Generative Artificial Intelligence (GenAI), AI Algorithms, Convex Optimization, Reinforcement Learning from Human Feedback (RLHF), Deep Reinforcement Learning, Reinforcement Learning, Deep Learning, Machine Learning, Artificial Intelligence (AI), Artificial Neural Networks (ANN), ChatGPT

Technical Research Fellow

2021 - 2024
US Naval Nuclear Lab & Stanford Wehab Lab
  • Developed quantum circuits for solving elliptic PDEs. Utilized Qiskit on IonQ, Quantinuum, and IBM quantum hardware to provide insights into quantum hardware performance. By appointment through the foundation: Womanium Quantum.
  • Led AI-driven mental health intervention research, using Shapley values and FDR for personalized mental health models. Earned honorable mention at CHI20 for innovation in AI mental health.
  • Authored IEEE publication celebrating our applied quantum computing research. Additionally, I published a second IEEE paper on artificial intelligence (AI) applied to mental health (panic attacks, particularly).
  • Published an award-winning conference paper at CHI20 on the subject of AI for preventing panic attacks. I was personally interviewed on this subject by the Stanford Daily.
Technologies: Quantum Computing, Quantum Circuits, Quantum Machine Learning, Quantum Approximate Optimization Algorithm (QAOA), Variational Quantum Eigensolver (VQE), Variational Quantum Linear Solver (VQLS), QML, Artificial Intelligence (AI), Explainable Artificial Intelligence (XAI), Bayesian Machine Learning, Bayesian Networks

Director | Data Science

2022 - 2023
PepsiCo
  • Designed, fundraised for, and delivered three internal software products. Each product was measured and demonstrably increased brand reach and consumer product sales.
  • Coordinated across data engineering, product management, marketing brand directors, and internal customer success managers to ensure smooth maintenance and operations of all three delivered software products.
  • Restructured three geographically siloed teams and performance-managed them. Retention was 90% throughout the restructuring and stayed at that level throughout my engagement.
  • Defined and hosted tech talks and tech demo presentations on an alternating biweekly cadence. Provided stakeholders with visibility into the individual contributors of my group. Ensured cross-functional collaboration with other groups.
Technologies: Data Science, Leadership, Machine Learning, Software Architecture, Performance Management, Software Design, Deep Learning, Spark ML, Stakeholder Management, Product Management, Feature Prioritization, Project Budget Management, Annual Budgets, Budget Management, Artificial Intelligence (AI), Product Planning, Product Roadmaps, Unsupervised Learning, Keras, TensorFlow, PyTorch, MXNet, Docker, Flask, FastAPI, Behavioral Science, Experimental Design, Bayesian Inference & Modeling, Docker Compose, Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, IT Consulting, Data, Data Visualization, Market Research, Business Consulting, Data Scientist, AI Programming, Neural Networks, Machine Learning Operations (MLOps), Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, AI Model Training, Technical Leadership, Consulting, Mentorship & Coaching, Forecasting, Scikit-learn, NoSQL, OpenAI, Data Modeling, Generative Artificial Intelligence (GenAI), LoRa, Hugging Face, Consumer Packaged Goods (CPG), SQL, XGBoost, LightGBM, Databricks, DevOps, AI Consulting, AI Design, Large-scale Production Deployments, API Integration, Workforce Management (WFM), Automation

Consultant | Artificial Intelligence

2021 - 2021
Antares Audio Technologies
  • Designed, implemented, and deployed two deep learning sequence models. Both models leveraged state-of-the-art (SOTA) mechanisms, such as shift operations as replacements for convolutions and masked self-attention as replacements for recurrence.
  • Ensured each of the two sequence models was interpretable and explainable to my two business stakeholders; the CMO and the COO.
  • Coordinated with their software engineers how to integrate the two sequence models into their automated business processing pipelines.
  • Coached the CMO and the COO on how to best consume the forecasts and the automatically generated explanations for their downstream automated business processes.
Technologies: Sequence Models, Explainable Artificial Intelligence (XAI), Deep Learning, Artificial Intelligence (AI), Time Series Analysis, Machine Learning, Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, IT Consulting, Data, Data Visualization, Data Scientist, AI Programming, Generative Pre-trained Transformers (GPT), Large Language Models (LLMs), Neural Networks, Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, Audio, AI Model Training, GitLab, Technical Leadership, Consulting, Forecasting, Scikit-learn, OpenAI, Data Modeling, Natural Language Processing (NLP), Generative Artificial Intelligence (GenAI), LoRa, Hugging Face, SQL, XGBoost, LightGBM, Databricks, AI Consulting, AI Design, Automation

Director | Data Science

2019 - 2020
Booster
  • Built their data team of four engineers. My team was horizontally integrated and served the needs of all other business units. My three business stakeholders were the CFO, the CMO, and the CPO.
  • Oversaw the creation of the company's business intelligence queries and dashboards and managed its quality assurance.
  • Participated in the design, implementation, and deployment of their demand forecasting systems for their B2B and B2C customer segments.
  • Participated in the experimental design and analysis for pricing elasticity and demand response to pricing changes.
  • Participated in the experimental design and analysis for marketing conversion and customer engagement with referral programs.
  • Performed supply-side pricing analysis and informed executive stakeholders throughout multiple contract negotiations with Fortune 100 enterprises.
  • Analyzed the quality and the efficiency gains of their proprietary vehicle routing software.
  • Managed the data science support of their engineering efforts, including data modeling and analytics for their proprietary IoT platform.
Technologies: Time Series Analysis, Sequence Models, Time Series, Demand Planning, Pricing Models, Pricing, Business Intelligence (BI), Experimental Research, Internet of Things (IoT), Industrial Internet of Things (IIoT), Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, Data, Data Visualization, Market Research, Business Consulting, Data Scientist, AI Programming, Neural Networks, Machine Learning Operations (MLOps), Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, AI Model Training, Technical Leadership, Mentorship & Coaching, Forecasting, Scikit-learn, NoSQL, Data Modeling, SQL, XGBoost, LightGBM, Databricks, AI Consulting, AI Design, Large-scale Production Deployments, Software Architecture, API Integration, Workforce Management (WFM), Automation

Consultant | Artificial Intelligence

2018 - 2020
Stanford University
  • Co-developed novel AI applications for personalized behavioral health treatments.
  • Took part in the work that was published in the peer-reviewed IEEE scientific journal.
  • Developed an AI system for personalized behavioral health treatment that was awarded an honorable mention in the peer-reviewed ACM CHI20 scientific conference.
Technologies: Health, Psychology & Mental Health, Cognitive Behavioral Therapy (CBT), Behavioral Science, Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, Data, Data Visualization, Market Research, Data Scientist, AI Programming, Neural Networks, Modeling, R, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, Audio, AI Model Training, GitLab, Technical Leadership, Consulting, Mentorship & Coaching, Forecasting, Scikit-learn, Data Modeling, XGBoost, LightGBM, Databricks, AI Consulting, AI Design, Automation

Manager | Data Science

2017 - 2019
Autodesk
  • Managed six machine learning engineers to create, deploy, and maintain three production systems in an Apache Spark powered data platform.
  • Participated in the design, implementation, and deployment of their enterprise's customer churn rate forecaster.
  • Contributed to the design, implementation, and deployment of the enterprise's sales recommendation system.
  • Collaborated with the design, implementation, and deployment of the enterprise's software piracy detection system.
Technologies: Generative Design, Generative Adversarial Networks (GANs), Deep Learning, Machine Learning, Explainable Artificial Intelligence (XAI), Adversarial Machine Learning, Recommendation Systems, Spark ML, Spark SQL, Apache Spark, Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, Data, Data Visualization, Business Consulting, Data Scientist, AI Programming, Physics, Generative Pre-trained Transformers (GPT), Large Language Models (LLMs), Neural Networks, Machine Learning Operations (MLOps), Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, AI Model Training, Technical Leadership, Mentorship & Coaching, Forecasting, Scikit-learn, NoSQL, Data Modeling, Generative Artificial Intelligence (GenAI), SQL, XGBoost, LightGBM, Databricks, AI Design, Large-scale Production Deployments, Software Architecture, API Integration, Workforce Management (WFM), Automation

Consultant | Artificial Intelligence

2017 - 2017
Qubole
  • Defined machine learning consulting offers for their B2B professional services division. My business' stakeholder was their senior vice president of professional services.
  • Implemented, deployed, and marketed two deep learning solutions for drug discovery. These solutions helped them land their first contracts with publicly listed pharmaceutical companies.
  • Improved their platform by demonstrating artificial intelligence solutions at large conferences, such as Strata and PyData.
  • Worked on the platform by accurately predicting trends at the intersection of big data and AI and ML during interviews with CIO magazine and Forbes.
  • Helped them win contracts with publicly listed banking and finance companies by providing deep learning and machine learning consultations with prospect clients.
Technologies: Spark ML, Spark SQL, Apache Spark, Artificial Neural Networks (ANN), Keras, TensorFlow, Big Data, BigDL, Amazon SageMaker, Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, Data, Data Visualization, Market Research, Data Scientist, AI Programming, Neural Networks, Machine Learning Operations (MLOps), Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, AI Model Training, GitLab, Technical Leadership, Consulting, Forecasting, Scikit-learn, NoSQL, Data Modeling, Generative Artificial Intelligence (GenAI), XGBoost, LightGBM, Databricks, AI Consulting, AI Design, Large-scale Production Deployments, Automation

Experience

Artificial Intelligence for Behavioral Therapy

https://www.stanforddaily.com/2019/10/21/researchers-work-on-device-to-help-individuals-with-autism-handle-stress/
Innovated in applied psychology by leveraging interpretable and explainable AI and ML to tackle emotional dysregulation, such as anxiety and panic attacks.

Our work was featured in Stanford magazine. Feel free to read the interview in the link provided.

Our work was also awarded at the peer-reviewed scientific conference ACM CHI20, and the US government granted us additional funding to continue our efforts.

Said grant money helped us develop the project further and subsequently publish in the peer-reviewed scientific journal IEEE.

Quantum Mini Apps for Engineering Applications: A Case Study

https://arxiv.org/abs/2411.12920
Developed quantum circuits for solving elliptic PDEs. Utilized Qiskit on IonQ, Quantinuum, and IBM Quantum hardware to provide insights into quantum hardware performance.

Acted as the lead author on an IEEE publication celebrating our work at the US Naval Nuclear Laboratory (by appointment through the foundation: Womanium Quantum).

Education

2015 - 2017

Master's Degree in Statistics

Stanford University - Stanford, California, USA

2008 - 2011

Bachelor's Degree in Computer Science

University of California, Berkeley - Berkeley, California, USA

2008 - 2011

Bachelor's Degree in Cognitive Science

University of California, Berkeley - Berkeley, California, USA

Certifications

DECEMBER 2024 - PRESENT

Generative AI with Large Language Models

Coursera

DECEMBER 2024 - PRESENT

Machine Learning in Production

Coursera

Skills

Libraries/APIs

Spark ML, Keras, Scikit-learn, XGBoost, TensorFlow, BigDL, PyTorch, JAX, LSTM

Tools

ChatGPT, GitLab, Spark SQL, Amazon SageMaker, Docker Compose

Languages

Python 3, Python, Julia, R, SQL, Scala, Java, JavaScript, TypeScript, Bash Script, QML

Frameworks

LightGBM, Apache Spark, MXNet, Flask, Multi-armed Bandits (MABs)

Paradigms

Automation, DevOps, Business Intelligence (BI)

Platforms

Jupyter Notebook, Databricks, Linux, Docker

Storage

NoSQL

Other

Deep Learning, Experimental Design, Natural Language Processing (NLP), Artificial Intelligence (AI), Machine Learning, Data Science, Explainable Artificial Intelligence (XAI), Causal Inference, Artificial Neural Networks (ANN), Recommendation Systems, Time Series Analysis, Sequence Models, Time Series, Generative Pre-trained Transformers (GPT), Probability Theory, Predictive Modeling, Predictive Analytics, Data Analytics, Large Data Sets, Data, Data Visualization, Market Research, Algorithms, Data Scientist, AI Programming, Large Language Models (LLMs), Neural Networks, Machine Learning Operations (MLOps), Modeling, Statistical Analysis, Regression Modeling, Statistical Modeling, Data Analysis, AI Model Training, Technical Leadership, Consulting, Mentorship & Coaching, Forecasting, OpenAI, Data Modeling, Generative Artificial Intelligence (GenAI), LoRa, Hugging Face, AI Consulting, AI Design, Large-scale Production Deployments, Cognitive Computing, Cognitive Psychology, Cognitive Science, Experimental Research, Natural Language Understanding (NLU), Stochastic Differential Equations, Stochastic Modeling, Deep Reinforcement Learning, Reinforcement Learning, Meta-learning, Multi-task Learning, Psychology & Mental Health, Cognitive Behavioral Therapy (CBT), Behavioral Science, Software Architecture, IT Consulting, Business Consulting, Computational Algorithms, Physics, Audio, Finance, Consumer Packaged Goods (CPG), API Integration, Workforce Management (WFM), LangChain, Computer Vision, Machine Vision, Probabilistic Graphical Models, Generative Design, Generative Adversarial Networks (GANs), Adversarial Machine Learning, Demand Planning, Pricing Models, Pricing, Internet of Things (IoT), Industrial Internet of Things (IIoT), Big Data, Health, Leadership, Performance Management, Software Design, Stakeholder Management, Product Management, Feature Prioritization, Project Budget Management, Annual Budgets, Budget Management, Product Planning, Product Roadmaps, Unsupervised Learning, FastAPI, Bayesian Inference & Modeling, Quantum Computing, Quantum Circuits, Quantum Machine Learning, Quantum Approximate Optimization Algorithm (QAOA), Variational Quantum Eigensolver (VQE), Variational Quantum Linear Solver (VQLS), Quantum Information, Transfer Learning, AI Agents, Light LLMs, Open-source LLMs, Contextual Bandits, Non-generative AI, Transformer Models, Long Short-term Memory (LSTM), LSTM Networks, Gated Recurrent Unit (GRU), Recurrent Neural Networks (RNNs), AI Algorithms, Convex Optimization, Reinforcement Learning from Human Feedback (RLHF), Bayesian Machine Learning, Bayesian Networks, Llama, Distributed Training, Parallelism

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.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

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