Hi!👋 I'm Isaac.​

My focus is on making AI systems scalable and maintainable. Currently I'm a Staff Machine Learning Engineer at Foam.

Previously I've worked on building ML-powered product features in the customer support domain (reasoning, query understanding & rewrite, query intents, query routing), and multi-modal search systems for product discovery. I have spoken at various Python & ML conferences, and meet ups in Europe. My stack includes Python, Docker, Kubernetes, PostgreSQL, and Go.

My background is in Aerospace Engineering and Machine Learning and I hold undergraduate (B.A.Sc in EngSci) and graduate (M.A.Sc) degrees from the University of Toronto.

I work remotely in Europe, currently based in Tallinn, Estonia. In my spare time, I try to contribute to open source projects (e.g. MTEB), see the world, and stay active. I used to race triathlons actively (a sneak peek here and here). These days I'm more into cycling, running, bouldering, and hiking.

Papers​

• Adnan El Assadi, Roman Solomatin, Isaac Chung, Chenghao Xiao, et al. MVEB: Massive Video Embedding Benchmark. The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2026.
• Adnan El Assadi, Isaac Chung, Roman Solomatin et al. MAEB: Massive Audio Embedding Benchmark. The Fortieth Annual Conference on Neural Information Processing Systems Evaluations and Datasets Track, 2026.
• Sahil Verma, ..., Roman Solomatin, Isaac Chung, Tom Aarsen, et al. RTEB: An Overfitting-Resistant Benchmark for Embedding Model Evaluation. The Fortieth Annual Conference on Neural Information Processing Systems Evaluations and Datasets Track, 2026.
• Adnan El Assadi, Isaac Chung, Roman Solomatin, Niklas Muennighoff, Kenneth Enevoldsen. HUME: Measuring the Human-Model Performance Gap in Text Embedding Tasks. The Fourteenth International Conference on Learning Representation, 2026.
• Isaac Chung, Linda Freienthal. Cross-Lingual Stability of LLM Judges Under Controlled Generation: Evidence from Finno-Ugric Languages. First Workshop on Multilingual Multicultural Evaluation, co-located with EACL 2026.
• Isaac Chung, Imene Kerboua and Márton Kardos and Roman Solomatin and Kenneth Enevoldsen. Maintaining MTEB: Towards Long Term Usability and Reproducibility of Embedding Benchmarks. Championing Open-source DEvelopment in ML Workshop @ ICML 2025
• Chenghao Xiao, Isaac Chung, Imene Kerboua, Jamie Stirling et al. MIEB: Massive Image Embedding Benchmark. International Conference on Computer Vision (ICCV) 2025.
• Kenneth Enevoldsen, Isaac Chung, Imene Kerboua, Márton Kardos et al. MMTEB: Massive Multilingual Text Embedding Benchmark. The Thirteenth International Conference on Learning Representation, 2025.
• Isaac Chung, Phat Vo, Arman C. Kizilkale, and Aaron Reite. Efficient In-Domain Question Answering for Resource-Constrained Environments. arXiv:2409.17648, 2024.
• Elizaveta Korotkova and Isaac Chung. Beyond Toxic: Toxicity Detection Datasets are Not Enough for Brand Safety. arXiv:2303.15110, 2023.

Talks and Conferences​

• 2026 PyCon Italia (Bologna, Italy): Reproducibility in Embedding Benchmarks | Video | Slides
• 2025 Data Science Summit (online): Parametrized Synthetic Conversations for LLM Evaluation and Enablement | Slides
• 2025 PyCon Wrocław (Wrocław, Poland): How AsyncIO Saved Our LLM production service
• 2025 Berlin Buzzwords (Berlin, Germany): Reproducibility in Embedding Benchmarks | Video | Slides
• 2025 PyData London (London, UK): Reproducibility in Embedding Benchmarks | Slides
• 2024 Swiss Python Summit (Zurich, Switzerland): Prototype to Production for RAG applications | Video | Slides
• 2024 Data Makers Fest (Porto, Portugal): Prototype to Production for RAG applications | Slides
• 2024 PyCon PL (Gliwice, Poland): Transcend the Knowledge Barriers in RAG | Slides
• 2024 PyCon LT (Vilnius, Lithuania): Speed up open source LLM-serving with llama-cpp-python | Video | Slides | Github
• 2024 PyCon LT (Vilnius, Lithuania): Transcend the Knowledge Barriers in RAG | Video | Slides
• 2023 TD Lab Live AI Talk (Remote): Beyond Llama2: Future Trends and Challenges with LLMs | Video | Slides
• 2023 EstoniAI Meetup Vol. 5 (Tallinn, Estonia): Panel Discussion on Recap AI developments and future trends
• 2022 ECIR Industry Day (Stavanger, Norway): Scaling Cross-Domain Content-Based Image Retrieval for E-commerce Snap and Search Application. Talk not recorded.
• 2021 Clarifai Perceive Conference (Remote): Automating Data Labeling for Deep Learning - AI-Automated Data Labeling | Video | Slides

Projects​

• Organizer @ PyData Tallinn
• Maintainer @ Massive Text Embedding Benchmark (MTEB)
• Strava Kudos Bot: https://github.com/isaac-chung/strava-kudos

Blogs​

• I log my learnings on Generative AI/ML in a blog and try to keep it within a 3-5min read
• Neptune LLMOps Blog on building RAG systems
• I have also written a few blogs for Clarifai.
  • Here are a few recent examples:

    • Supercharge your LLM via Retrieval Augmented Fine-tuning
    • The Landscape of Multimodal Evaluation Benchmarks
    • Do LLMs Reign Supreme In Few-Shot NER? Part III
    • Do LLMs Reign Supreme In Few-Shot NER? Part II
    • Do LLMs Reign Supreme In Few-Shot NER?
    • Multi-modal Moderation
    • A Comprehensive Guide To Vector Search

Consulting​

I'm open to provide consulting services in ML/AI. Send me an email, or reach out over LinkedIn.