Pura

Senior Data Engineer

Dutch Engineer

I can write blog posts, LinkedIn posts, demos, or share advice about your data or machine learning infrastructure.

Founder

EcZachly Inc

Instructor and Lead Teacher Assistant

Zwift

Build long-term roadmaps for machine learning platform and code best practices for the data science team
Orchestrated the implementation of DBT to build data lineage and shorten debugging time by 30%
Deployed XGBoost models to identify customers who will might or not subscribe in free trials. 
Lead a team to provide the best short-term and long-term solutions to reduce service downtime by 50%
Own e-commerce datasets to lower handle data quality, data contracts, and resolve pipelines downtime
Tools: AWS, Kafka, Python, Docker, Kubernetes, Scala, Spark, SQL, Airflow, Jenkins,  DBT

Senior Data and ML Platform Engineer

Clinical Trial Media

Documented technical specifications of workflows to be presented to stakeholders and executive team
Automated and optimized existing workflows using Prefect and dbt to adhere to best practices
Tools: DBT, Prefect, Azure, Azure Data Factory, Python, SQL, Tableau Server, Docker

Qumulo

Lead migration of on-premise pipelines to cloud capable of handling 210+ GB per day
Produce key metrics to counsel sales and engineering and drive product feature decisions
Integrate linters, tests, and CICD to improve Data Science code quality 
Tools: AWS, Redshift, S3, Postgresql, Pyspark, Fargate, Python, Gitlab CICD, SQL, MongoDB, Tableau

Data Scientist

Blueprint Technologies

Created 30,000 customers’ metric sheets that calculated employees behavior for 1.5 months before and 1.5 months after covid and presented to the c-executive suite of the F500 company 
Developed daily pipelines with >500 TB that build employee behavior metric sheets on request
Productionalize Decision Tree and Logistic Regression to prioritize likely buyers
Optimize SQL and PowerShell scripts to reduce running script from 3 days to 15 minutes.

Tools: Azure Databricks, Python, Plotly, PowerShell, C#, COSMOS, SQL

Data Engineer

Redapt Inc

Migrating Hadoop cluster on VMs to Google Cloud Platform's Dataproc cluster
Importing, cleaning, and visualizing data in Power BI dashboards
Building a storage infrastructure system to identify frequent patterns and errors 
Incorporating Machine Learning algorithms to recommend new items for over a million items and for several thousands of customers. 
Tools: Hadoop, Kafka, Airflow, Spark, Power BI, Azure Databricks, Azure ML Studio, Azure Blob Storage

Apex Learning

Build a python script to obtain Zendesk's data via its RESTful API for marketing reports
Redesigned monthly and yearly reports by incorporating Tableau Software to evaluate tutors’ performance 
Implement Salesforce.com to create an easily accessible school’s credit acceptance database for marketing campaigns 
Support middle and high school students’ through the Apex Learning Virtual School curriculum

Online Tutoring and Data Analysis

Projects

ILest: Ionic Liquid estimation Tool
Team project, Implemented LASSO, SVR, and MLP-Regressor on 10,000 data points to estimate the desired properties of ionic liquid binary mixtures for electric battery applications
Tools: scikit-learn, Pandas, RDKit, Matplotlib

Data-Driven Design of Perovskite
Team Project, Carried out LASSO, and supervised neural networks on a reduced data set to identify the optimal combination of elements in the hybrid organic perovskites (crystals)
Tools: scikit-learn, Pandas, RDKit, Matplotlib

Stack Overflow: Suggest a Tag for My Post Please 
Queried 150k posts and users details to determine tags needed based upon a posts' body or score.
Tools: PySpark, Neo4j, Plotly, Python, Cypher, SQL

How Much Will Your Videogame Sell For? 
Web scraping video game tweets and statistics from VGChartz to predict their unit sales using machine learning
Tools: BeautifulSoup, Matploblib, multiprocessing, requests, plotly

Twitter Scraping Tool
Web-scraped old tweets from the twitter page to eliminate the usage of the Twitter Official API and decrease the time needed by 60%. Tools: pyquery, requests, datetime

Senior Data & ML Platform Engineer | dutchengineer.substack.com | Host of Ask A Data Mentor Podcast

Built long-term roadmaps for data platform and machine learning platform
Experience in designing, analyzing, and visualizing large (>500 TB) datasets 
Build and manage multiple ETL infrastructures to identify text patterns and detect network and hardware anomalies
Prepare critical metrics to guide product decisions in people, healthcare, computer games, and chemical analytics
Multilingual (Dutch, English, Spanish)

Keywords: Airflow, AWS (Cloudwatch,  ECS, EKS, Fargate, Glue, Redshift, Kinesis, Databricks, Postgresql, S3), Azure (Azure Functions, Azure ML Studio, Databricks, PowerBI, SQL Server), BentoML, C++, Datadog, Docker, ElasticSearch, Flink, Kafka, Keras, Kubernetes, Hive, High Performance Computing, Github Actions, Gitlab CICD,  Pagerduty, PowerBI,  Prefect, Python (DBT, Numpy, Pydantic, Sklearn,Scipy), Pyspark, Neural Network, Natural Language Processing, MLFlow, MongoDB, Recommendations, Sagemaker, Scala, SparkML, SQL, Tableau, Terraform

Sarah Floris, MS

👩🏻‍💻 Mikiko B.

Labelbox

Try Labelbox for all your unstructured labeling, automation, catalogue, fine-tuning & data evaluation needs!

Labelbox is a Data-Centric AI Platform focused on helping companies unlock their unstructured data (audio, video, text, geospatial, etc) to develop & train foundational models as well use the latest, cutting edge CV, NLP, & LLM  models to explore, optimize, prepare, & evaluate data sets for applications.

🔗 https://bit.ly/40wvulk

Head of AI Developer Relations

Featureform

During my time at Featureform, my main goals were to:

✔️ Develop FF as a thought leader in the MLOps space;
✔️ Raise awareness of the FF brand and connecting with practitioners by speaking at well-respected conferences and community meetups; 
✔️ Serve as the in-house ML expert and provide my insights and expertise to the product/dev and sales teams through internal education and demo assists (including developing notebooks & examples);
✔️ Craft the social media voice of FF.

Key Activities & Results

✅ Collaborated with design and marketing to produce high-quality, practitioner focused content about best practices in feature engineering and ML productionization. Notables include:

👉 “The Feature Engineering Guide”: A 40+ page introduction to the ML lifecycle, feature engineering, and the challenges of the “art” of data science and ML development.
👉 “Measuring Your ML Platform's North Star Metrics”: A 20+ page guide for ML platform teams and MLOps engineers on how to apply product management principles to thoughtfully navigate the build-buy decisions with regards to MLOps tooling. 

✅ Developed 4+ presentations on feature stores, MLOps, Gen-AI, data science for conferences & webinars like Convergence, O’Reilly Kubernetes Superstream, etc. Some notables: 

👉 DataEngBytes 2023: 3 cities in 2 weeks, 3 custom talks + 3 panels (featured international speaker) – Over 500+ data engineers attended my specific talks. 
👉 Data Council 2023: “The Fun-Sized MLOps Stack” was packed with standing room only, with some attendees saying it was so good, they’d forgotten I was “speaking from a vendor company”. 

✅ Helped author and promote tutorials and docs for dev (including the release notes and minor version announcements) to help highlight new features for the open-source repo, along with the social media announcements (ex: “Feature Store Tutorial: Feature Versioning 101”, “Featureform v0.🔟 Highlight 🔦: The New & Improved Dashboard”).

Head of MLOps

NVIDIA

Sharing industrial-strength insights, perspectives, & knowledge of ML, deep learning, distributed system design, & software engineering best practices with the broader data science community.

Contributing Writer

Intuit

Working with: GCP (BigQuery, AI Platform, Dataflow), Python, PyTest, Great Expectations, Docker, Kubernetes, Jenkins, Airflow, Github (Cloud, Actions, Packages), Cookiecutter, SQL, Bash, Terraform

🏆 Key Accomplishments: 
✔️ Selected for the first Global Mailchimp Engineering hiring committee as ML Ops & Data Engineering sole representative;
✔️ Created interviewing process & collaborated on developing system design interviews for ML Ops team;
✔️ Lead change management of major upstream templating command-line tool including updates to 50+ downstream repos, as well as managing merge conflicts, performing code reviews, & developing initial repo health checklist. Also shipped improvements to DS project templating tools including bugfixes & patches;
✔️ Provided initial health analysis of the portfolio of production ML projects; 
✔️ Won Mailchimp GPT-3 Hackathon & pitched idea to product;
✔️ Overhauled & upgraded documentation for ML Eng+Ops confluence space, including improvements to onboarding documentation & process;
✔️ Developed "Software Engineering Skills 101" learning path for Data Science team, including modules on testing, version control, & Python language fundamentals;

✍🏻 Notable Responsibilities: 
✔️ Maintenance of internal tooling support for Data Scientists, including fixing bugs & shipping new features
✔️ Participate in interviewing committee for MLOps, data engineering, & data science;
✔️ Help develop workshops & internal training on software engineering best practices for Data Science team, including testing; 
✔️ Mentor & manage ML Ops internship, project & sprint planning included;
✔️ Ran monthly Data Book Club for senior, staff+ engineers focused on systems design, data engineering, data platforms, GCP, & MLOps best practices;

Senior Software Engineer, MLOps

Mailchimp was acquired by Intuit in Nov 2021

Mission:
🙉    enable our data scientists to develop cool features;
🙈    empower Mailchimp customers to understand and utilize their data;
🙊    support the MLE team in efficient & scalable deployment, integration, monitoring, & testing.

Software Engineer, MLOps

Sidewalk

Worked on: Data Eng; ML Development & Deployment; Analytics; Experimentation; Creating NLP, CV, & Geospatial products.

Building a new SaaS platform for residential real estate.

Software Engineer, MLOps & Data

Teladoc Health

Worked with: Python, SQL, Jupyter, Pandas, Scikit-learn, Seaborn, Plotly, Scipy 

Collaborate closely with leaders at Livongo to deliver world-class decision support for growth marketing and helping increase Livongo's reach and impact for people living with chronic and behavioral conditions. 

Livongo was acquired in Oct 2020 by Teladoc.

Data Scientist, Growth

WalkMe™

Worked with: Python, Pandas, Scikit-learn, Seaborn, Plotly, Scipy, Spacy, Gensim, SQL, Tableau, Salesforce, Clari

Responsible for leading & executing advanced analytics & data science initiatives for the sales organization. Includes working with key stakeholders to determine appropriate ML projects, creating ML models, implementing & developing DS best practices, & serving as the key analytics partner to the sales org.

Senior Analyst, RevOps

Autodesk

Sole adoption team data science asset focused on understanding & driving user adoption of core products of construction software portfolio.

Sunrun

Sales & supply chain FP&A partner for largest residential solar company with special focus on advanced analytics, S&OP planning & forecasting, & Anaplan modeling.

Analyst, Finance

Digimarc Guardian

Work cross-functionally in order to methodically understand the needs of our primary customers and internal stakeholders in order to focus the organization in profitably creating, communicating and delivering superior value in the form of improved tools, service efficiency, and transparency into organizational performance.

BizOps Associate

🤖 {Helping MLOps teams build platforms} + {Data Scientists get models into production} 🤖

My personal mandate: 
⭐️ Develop ML platforms & MLOps systems that work, no matter the business or technology constraints;
⭐️ Contribute to the open-source MLOps ecosystem & continue to drive innovation;
⭐️ Create high-quality & thoughtful content that helps everyone in production ML become more productive, collaborative, & happier. 

As an MLOp practitioner & educator I do this by:  
✅ Hacking projects using best-of-breed tools & practices from software engineering, DevOps, data engineering, & machine learning using open-source & public cloud (GCP, AWS);
✅ Consulting with startups & enterprises (at all levels of Data & ML maturity) developing production ML pipeline, systems, & tooling; 
✅ Writing books, videos, courses, & workshops to teach MLOps best practices & architectures;
✅ Growing communities & facilitating dialogue through my youtube, substack, twitter, & twitch platforms. 

I've lead & executed key Data Science & MLOps initiatives at Mailchimp (Intuit), Teladoc, WalkMe, Autodesk, and Sunrun☀️ and partnered on content with companies like Nvidia, Bright Data, & HP. 

I've also mentored at bootcamps (Springboard & Data Science Dream Job) & spoken at a number of conferences & career panels (Lesbians Who Tech 🏳️🌈, General Assembly, etc) on breaking into data science & machine learning. 

In my free time I love doing CrossFit 🏋🏻, designing & making clothes & shoes 👟, boxing 🥊 , and spending time with my family & friends. Newest hobby also includes amateur birding. 🦅 🔭

📝 Substack: https://mikikobazeley.substack.com/
📹 Youtube: https://www.youtube.com/@Miki_ML

Also here:
👉🏻 LinkedIn: https://www.linkedin.com/in/mikikobazeley/
🐙🐈 Github: https://github.com/MMBazel
📱 Titok: https://www.tiktok.com/@mlops_miki
📷 IG: @miki_themleng
🐥 Twitter: https://twitter.com/BazeleyMikiko

📒 Remember when everyone hated Jupyter notebooks in production?

2015: Data scientists discovered Project Jupyter notebooks. Finally, they could iterate fast, visualize results, and share work effortlessly.

2016: Engineering teams revolted. "You can't put notebooks in production! Where's the version control? Testing? Deployment pipelines?"

Some teams tried anyway. Most failed spectacularly.

But here's what we learned: Notebooks weren't the enemy. They were a development tool being forced into the wrong role.

Fast forward to 2025:

We're watching the exact same tension play out with "vibe-coding"—using AI assistants (Claude, Cursor, Copilot) to rapidly prototype agent systems.

The Criticisms Sound Familiar:
❌ "You can't just prompt your way to production systems"
❌ "This isn't real engineering" 
❌ "What about testing? Monitoring? Security?"

And you know what? They're right. AND wrong.

Here's what actually matters:

Just like Jupyter notebooks found their place in the workflow (exploration → validated code → internal & external dynamic communication and MVPs for both technical and non-technical users), vibe-coding has its place:

✅ Use AI assistants to rapidly build POC after you've validated the problem
✅ Use structured frameworks (like canvases) to ensure you're building the right thing
✅ Use proper engineering practices when moving to production

The pattern I'm seeing in successful teams:
1. Canvas planning (structured thinking)
2. Vibe-coding for rapid POC (AI-assisted development) 
3. Production-grade engineering for scale (proper architecture)

Notebooks didn't replace production systems. They accelerated the path to them.

Vibe-coding won't replace software engineering. It accelerates the path to validated, production-ready agents.

The teams that win aren't the ones who reject new tools—they're the ones who understand when and how to use them.

And I can bet we'll experience something similar with MCP. 

Also: MongoDB's MCP server is available for use in Anthropic's Claude Desktop, Microsoft's VS Code, Cursor, Google's Gemini CLI, Windsurf, Augment Code

#aiagents #agentsinproduction #agentops #llmops #mlops #AIEngineering #AgenticAI #MCP #vibecoding

Staff Developer Advocate @ MongoDB | Gen-AI/ML Engineer | Powering Bleeding-Edge Agentic DevTool & Infra Teams

BazeleyMikiko - Vibe-coding agents

OpenAI just released its State of Enterprise AI report, and the adoption signals are striking.

A few highlights:
 • 75% of workers say AI improves the speed or quality of their work
 • Users report 40–60 minutes/day saved, with some saving far more
 • 75% say they can now perform tasks they couldn’t do before
 • Coding-related activity among non-technical roles is up 36%
 • International API usage has grown 70%+ over the last 6 months
 • The “AI divide” is widening—frontier workers send 6× more messages than the median, and frontier firms use 7× more GPT-driven workflows


The report makes one thing clear:
 AI is no longer experimental inside enterprises — it’s becoming infrastructure.

But here’s where my own experience with agent systems comes in:
 While adoption and productivity are rising fast, reliability remains the biggest barrier to deploying autonomous AI agents in production.

Most organizations I work with run into the same issues:
 • context drift and polluted memory
 • lost state across sessions
 • brittle prompt chains
 • unpredictable execution loops
(which can be exacerbated with multi-agent systems)

The result is a ceiling:
 Tools are scaling. Agents aren’t — yet.

This is why so many leading organizations are shifting toward what I call bounded autonomy:
 agents that operate independently within constraints, maintain state, and have observable decision paths.
 
It’s the pragmatic middle path between rigid deterministic workflows and fully autonomous systems.

As AI usage deepens across workflows (as the report shows), the next frontier isn’t just more adoption — it’s more reliable autonomy.

Curious to hear from folks building this internally:
Does this report match what you’re seeing? And where are your biggest gaps — adoption, reliability, or workflow integration?

taplio

The Battle

Sarah Floris, MS

👩🏻‍💻 Mikiko B.

Followers

Posts

Engagements

Likes

Performance Score👩🏻‍💻 Mikiko B.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Sarah Floris, MS

👩🏻‍💻 Mikiko B.

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}👩🏻‍💻 Mikiko B.'s LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance Score👩🏻‍💻 Mikiko B.'s LinkedIn Performance