Posts

Fine-tuning, distillation, and quantization sound similar but solve different problems. This LEGO explainer shows how.

Gave a talk on Engineering Open Models for Production alongside Dylan Bristot.

Delivered a talk ‘Coding with open models’.

Hosted a live, developer-focused office hours session on Nebius Token Factory, joined by Chris Alexiuk from NVIDIA to discuss Nemotron models.

Presented a session on coding with open models and coding agents

Supported builders as a technical mentor during the hackathon

Gave a talk on Coding with Open Models.

Organized and hosted a joint Webinar with Kimi team on the release of Kimi K3 model on Nebius Token Factory.

Demoed Token Factory and engaged with developers and attendees at the Nebius booth.

Delivered the talk: Adding real time intelligence to OpenClaw using Tavily and Token Factory.

Ran a Token Factory workshop at Berkeley SkyDeck (Batch 22)

Supported Butter Base Hackathon in Sunnyvale, CA.

Supported the event through Token Factory, and I gave a kickoff talk on Token Factory and open models.

Co-hosted a Builders & Brews meetup in London with Tavily, supporting developers with Token Factory credits and technical Q&A.

Co-presented a session on ‘Adding real time intelligence to OpenClaw using Tavily and Token Factory’.

Ran an office hour at Berkeley SkyDeck.

Demo: “Deep agents + Tavily + Nebius Token Factory”

Gave a talk ‘Engineering Open Models for Production’ alongside Dylan Bristot.

Talk: Building with open models.

Talk: Engineering Open Models for Production.

Hosted a webinar on turning production LLM logs into better models - observability and evaluation techniques for inference workloads.

Created and ran demos on Token Factory at the GTC booth.

Very first event on the ‘Nebius Build’ series - a pre-GTC workshop and hackathon in San Francisco. Organized / presented / supported the event.

Webinar on post-training techniques for running open-source LLMs in production.

Supported the NVIDIA Cosmos Hackathon - building on NVIDIA Cosmos models, alongside Nebius. Provided platform credits and supported participants.

Linux Mint 22 installation with best practices for disk partitioning and painless upgrades

Webinar on the production playbook for open-source LLMs, covering fine-tuning, distillation, and deployment techniques.

Talk: Engineering Open Models for Production.

Hosted the first Nebius Builder Hour, a one-hour online session for developers building with Nebius Token Factory. Covered what’s new in Token Factory, live builder demos, and open discussion.

Ran two workshops on Allycat and another on an open-source RAG pipeline

Ran a workshop on Allycat at Tech Equity AI Summit

Meld is a great open source diff tool for MacOS

supported the hackathon, served as judge

Gave a talk at the Berkeley AgentX/AgentBeats Competition, walking through Nebius Token Factory, demoing a CrewAI research agent, and comparing models.

Attended the NextBio Hackathon at UCSF in San Francisco, gave a talk on open-source models, and demoed AI Studio. Onboarded attendees to AI Studio.

How I migrated my site from Pelican to Hugo using Claude Code

Moving the site to hugo

Actionable strategies for running a smooth and successful unconference

My personal review of some pelican themes

moving the site to pelican

This hackathon focused on open source AI tools. Organized by lablab.ai.

My talk: Preparing data for LLM applications using Data Prep Kit.

Talk: Preparing Data for LLM Applications Using Data Prep Kit.

Numhack is focused on positive social impact using AI. Organized by Pydata and Numfocus.org.

Ran a workshop on preparing data for LLMs using data prep kit.

Talk: Preparing Data for LLM Applications Using Data Prep Kit.

Talk: Streamlining Data Prep for LLMs with Data Prep Kit.

This workshop focused on developing LLM applications, covering tokenizers, embeddings, vector search, local LLMs, and RAG.

Ran a hands-on workshop introducing Data Prep Kit. This is the first of the Data Prep Kit series.

Developing RAG applications using Mistral LLM.

Vector Search + LLMs with Dave Nielsen of MongoDB.

Talk: Vector Search with MongoDB + LLMs with Dave Nielsen of MongoDB.

Gave a talk about vector search applications and helped hackathon participants.

Ran a workshop for Data Science in Python.

Talk: Apache Spark - a hands on workshop.

Presented BigDL framework.

Presented my work on Intel’s BigDL framework.