Ivan is currently unavailable
Ivan Itzcovich
Verified Expert in Engineering
Artificial Intelligence Developer
Buenos Aires, Argentina
Toptal member since September 15, 2021
Bio
From Microsoft headquarters to co-founding its own startup, Ivan has worked in diverse scenarios where he learned the tradeoffs between stability and speed. He understands how to use appropriate engineering practices depending on each company's technical needs, whether building robustness for production or hitting the gas for validation. Ivan has strong machine learning experience in production, identifying its challenges as engineering-first quests.
Portfolio
Cased
Python, Large Language Models (LLMs), Artificial Intelligence (AI), Django...
Gantry
Docker, Kubernetes, Tilt, NumPy, REST APIs, Terraform, Druid.io, PostgreSQL...
ASAPP
Python, Apache Airflow, TensorFlow, PyTorch, Docker, Kubernetes, GPU Computing...
Experience
- Best Practices - 10 years
- Python - 10 years
- Data Structures - 10 years
- Algorithms - 10 years
- Machine Learning - 7 years
- Deep Learning - 7 years
- Artificial Intelligence (AI) - 7 years
- Large Language Models (LLMs) - 3 years
Work Experience
Lead AI Engineer
2023 - 2025
Cased
- Led a team of three engineers trying to find product market fit in the DevOps space with a large language model (LLM)-powered platform.
- Worked with SOTA LLMs to replicate tasks for DevOps engineers. The goal was to build "Cursor" for DevOps.
- Developed an LLM-powered chatbot using OpenAI's Assistants API, with deep integrations into DevOps tools like GitHub and AWS, enabling infrastructure management directly from Slack—an early prototype of a Mission Control Platform (MCP) client.
- Implemented an LLM-based linter to optimize the Terraform infrastructure. The system would detect security issues and cost optimizations to help maintain the Terraform repo.
- Implemented an LLM-based incident management system that suggested code changes to solve bugs in production, integrated with Sentry and GitHub.
Technologies: Python, Large Language Models (LLMs), Artificial Intelligence (AI), Django, Terraform, Amazon Web Services (AWS), GitHub, Slack, litellm, HTMX, OpenAI API, Anthropic, OpenAI, Full-stack Development
Founding Machine Learning Engineer
2021 - 2023
Gantry
- Developed and maintained the integration between the Gantry system and multiple LLM vendors, including OpenAI, Cohere, and Anthropic APIs.
- Designed and implemented a high-level abstraction for LLM prompt configurations to enable customers to do prompt engineering with their LLM.
- Developed and maintained the Python SDK client for data ingestion. Almost all customer production data was ingested into the platform through this SDK, meaning this software was optimized for high performance.
Technologies: Docker, Kubernetes, Tilt, NumPy, REST APIs, Terraform, Druid.io, PostgreSQL, APIs, Machine Learning, Natural Language Processing (NLP), Startups, OpenAI API, API Integration, Software Architecture, Large Language Models (LLMs), OpenAI, Full-stack Development
Lead Machine Learning Engineer
2018 - 2021
ASAPP
- Developed a custom deep learning training framework on top of PyTorch called Flambé. I presented it at the ACL conference in 2019.
- Built the entire training pipeline for our machine-learning models using Airflow. The implemented DAGs covered data ingestion, training, model storage, and metrics computation, making the process reliable and reproducible.
- Implemented a model storage and metrics platform for machine learning model comparison.
- Reimplemented our core ML API server, where all machine learning models were deployed. This was designed for high availability in massive traffic loads and implemented with low-level async Python support.
- Interviewed 50+ candidates to grow our engineering hub in Argentina from two people to 80+ people. Evaluated technical skills for different seniority roles and also team fit aspects for culture building.
- Implemented text classification models for intent detection using a supervised learning approach with prototypical architectures that enabled dynamic categories with very few data samples.
- Implemented a training pipeline for sentiment analysis in chat conversations between customers and call center agents to predict customer satisfaction.
Technologies: Python, Apache Airflow, TensorFlow, PyTorch, Docker, Kubernetes, GPU Computing, Aiohttp, Best Practices, Data Structures, Artificial Intelligence (AI), Linux, MacOS, Machine Learning, Deep Learning, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Neural Networks, Amazon Web Services (AWS), Deep Neural Networks (DNNs), Asyncio, REST APIs, Sentiment Analysis, Text Classification, Artificial Neural Networks (ANN), Recurrent Neural Networks (RNNs), Classification Algorithms, Regression Modeling, Pandas, Text-to-Speech (TTS), APIs, Startups, NVIDIA CUDA, Software Architecture
Data Scientist
2018 - 2018
Globant
- Implemented forecasting models for theft detection on retail companies.
- Developed a deep learning model for Argentinian plates detection using synthetic images.
- Built a pipeline for synthetic data generation for license plates.
Technologies: Python, TensorFlow, Data Science, Data Engineering, Data Synthesis, Artificial Intelligence (AI), Machine Learning, Deep Learning, Computer Vision, Neural Networks, Deep Neural Networks (DNNs), Object Detection, Artificial Neural Networks (ANN), Convolutional Neural Networks (CNNs), Classification Algorithms, OpenCV, Image Recognition, NVIDIA CUDA
Assistant Professor
2017 - 2018
Instituto Tecnológico de Buenos Aires
- Led the practical part of the "Data Structures and Algorithms" course in a 4-hour weekly class for approximately 30 software engineering students.
- Taught how to implement complex data structures and algorithms to consume them—mainly trees and graphs.
- Created all exams for the class and a 3-hour open-book coding challenge with three exercises.
Technologies: Java, Algorithms, Data Structures
Machine Learning Founding Engineer
2017 - 2017
Entelai
- Started the engineering efforts of a founding startup from scratch.
- Built a pipeline for brain imaging disease detection from data ingestion to visualization.
- Developed cutting-edge machine learning models for 3D medical imaging. More information on the Deep Brain project can be found under the experience section.
Technologies: Python, Flask, Docker, Machine Learning, Data Engineering, Artificial Intelligence (AI), Deep Learning, Computer Vision, Medical Imaging, Neural Networks, Deep Neural Networks (DNNs), Artificial Neural Networks (ANN), Convolutional Neural Networks (CNNs), Classification Algorithms, Medical Diagnostics, Startups, Health, NVIDIA CUDA, Software Architecture
Software Engineer Intern
2017 - 2017
Microsoft
- Developed a web application for internal use to manage release rollouts and throttling based on simple geographic rules.
- Tracked and reported bugs in their internal APIs, which were owned by a completely different team.
- Oversaw releases for Windows 10 updates worldwide using the created platform.
Technologies: .NET, JavaScript, APIs, C#, API Integration, Software Architecture
Experience
Lila: AI Testing Framework for Startups
I'm the founder of Lila, a QA AI agent for startups. Lila introduces a new way of declaring E2E tests on web applications using human instructions and uses LLMs to run them. Lila is a no-coding testing framework that allows you to test what your users do and not what you code.
Ivy: The Ivy Lee Method for Slack
I created this Slack app that implements the Ivy Lee method for Slack. This method is a productivity hack that has good results in remote teams. I implemented this so that members of a workspace could give visibility on their productivity to their team members.
Deep Brain
https://github.com/iitzco/deepbrainA machine-learning Python library for brain imaging processing. The idea of this library is to provide a basic set of validated and SoTA models for brain imaging processing (MRIs).
Currently, this project is on stand-by due to my lack of availability to continue its development.
Currently, this project is on stand-by due to my lack of availability to continue its development.
Faced
https://github.com/iitzco/facedA near-realtime open-source library for face detection. The goal was to implement a face detection mechanism that could run in simple hardware without needing a GPU.
Currently, this project is not being actively developed due to my lack of time, but it has been gathering positive feedback on the open-source community.
Currently, this project is not being actively developed due to my lack of time, but it has been gathering positive feedback on the open-source community.
Nalu Tech
An entirely digital product to build the best Google Docs experience ever made. From a deep integration with Google Drive official API to a desktop app, Nalu is an end-to-end product for busy people trying to have an awesome experience in Google Docs.
Education
2012 - 2017
Master's Degree in Software Engineering
Instituto Tecnológico de Buenos Aires (ITBA) - Buenos Aires, Argentina
Skills
Libraries/APIs
OpenAI API, TensorFlow, PyTorch, NumPy, Asyncio, REST APIs, Pandas, OpenCV, HTMX, Slack API, Bolt
Tools
PyPI, GitHub, Apache Airflow, Tilt, Terraform, Slack
Languages
Python, Java, JavaScript, C#
Paradigms
Best Practices
Platforms
MacOS, Linux, Docker, Kubernetes, Amazon Web Services (AWS), NVIDIA CUDA
Frameworks
Flask, Django, .NET, Electron
Storage
PostgreSQL, Druid.io, Redis, Graph Databases
Other
Artificial Intelligence (AI), Data Structures, Algorithms, Machine Learning, Computer Vision, Neural Networks, Deep Neural Networks (DNNs), Artificial Neural Networks (ANN), Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), APIs, Image Recognition, Startups, API Integration, Software Architecture, Large Language Models (LLMs), OpenAI, Full-stack Development, Data Engineering, Data Science, GPU Computing, Aiohttp, Deep Learning, Natural Language Processing (NLP), Sentiment Analysis, Text Classification, Object Detection, Classification Algorithms, Regression Modeling, Generative Pre-trained Transformers (GPT), Medical Diagnostics, Health, Meta Llama, Data Synthesis, Medical Imaging, Text-to-Speech (TTS), Web Scraping, litellm, Anthropic, AI Agents, Agentic AI, Browser Use, Open-source LLMs, Render, Slack App
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