Uploaded April 2026 | Updated September 2026, 3 weeks ago
From running one of the most aggressive internal AI rollouts in tech to building systems that simulate customers, optimize ML pipelines, and rethink search latency from first principles, Mikhail Parakhin is pushing Shopify far beyond “add AI to ecommerce.” In this episode, Shopify’s CTO joins swyx to unpack what actually changed inside the company after the December 2025 model inflection, why nearly everyone at Shopify now uses AI tools daily, and what happens when code generation outpaces the rest of the software delivery pipeline.
We go deep on Shopify’s internal AI stack: Tangle for reproducible ML and data workflows, Tangent for auto-research loops that keep iterating toward better results, SimGym for customer simulation and theme optimization, and Liquid AI for ultra-low-latency and long-context workloads. Mikhail also reflects on PR review in the age of coding agents, the future of CI/CD, Sidekick Pulse, UCP, hiring at Shopify, and the untold story behind Sydney at Microsoft.
We discuss:
• Why Shopify’s AI usage effectively reached company-wide adoption
• The December 2025 inflection when model quality suddenly changed behavior
• Why token budgets matter, but only if paired with strong critique and review loops
• The real bottleneck in AI coding: PR review, CI/CD, and deployment stability
• Why more AI-written code can still mean more bugs in production
• Tangle: Shopify’s reproducible workflow system for ML and data experimentation
• Tangent: auto-research loops for optimizing pipelines, prompts, throughput, and infra
• Why auto-research is democratizing ML experimentation beyond researchers
• SimGym: simulating customers using Shopify’s historical data and browser-based agents
• Why customer simulation is hard to copy without years of real merchant behavior
• HSTU, counterfactuals, and modeling merchants and buyers over time
• UCP and Shopify’s product catalog as runtime infrastructure for agents
• Why Shopify is using Liquid AI for low-latency search and long-context distillation
• Sidekick Pulse, recsys, and where Shopify’s AI platform is heading
• Hiring areas at Shopify across ML, data science, and distributed systems
• The backstory of Sydney, personality shaping, and the early Bing chatbot era
—
Mikhail Parakhin
• LinkedIn: linkedin.com/in/mikhail-parakhin
• X: https://x.com/MParakhin
Timestamps
00:00:00 Introduction
00:02:05 Mikhail Parakhin, Shopify, and Microsoft
00:03:20 Shopify’s AI Adoption and the December 2025 Inflection
00:10:07 Why Token Count Alone Is a Bad Metric
00:12:59 Why Shopify Built Its Own AI PR Review System
00:16:16 Why Git, PRs, and CI/CD May Break in the Agent Era
00:20:29 Tangle: Shopify’s System for Reproducible ML Workflows
00:28:19 Tangent: Auto Research for Optimization and Experimentation
00:34:58 The Limits of Auto Research
00:38:40 Why Tangle, Tangent, and SimGym Compound Together
00:39:24 SimGym: Simulating Customers with Shopify’s Historical Data
00:44:51 The Infra Behind SimGym
00:48:04 Why Shopify’s Data Gives SimGym a Moat
00:49:34 Counterfactuals, HSTU, and Modeling Merchant Trajectories
00:55:34 UCP, Shopify Catalog, and Identity Linking
00:57:11 Liquid AI: Why Shopify Uses Non-Transformer Models
01:01:17 Real Shopify Use Cases for Liquid
01:05:04 Can Liquid Scale Into a Frontier Model?
01:06:52 Sidekick Pulse, Recsys, and Shopify’s AI Stack
01:09:29 Hiring at Shopify: ML, Data Science, and Databases
01:10:43 Sydney at Bing: Personality Shaping and AI Character
01:13:32 Closing Thoughts
From running one of the most aggressive internal AI rollouts in tech to building systems that simulate customers, optimize ML pipelines, and rethink search latency from first principles, Mikhail Parakhin is pushing Shopify far beyond “add AI to ecommerce.” In this episode, Shopify’s CTO joins swyx to unpack what actually changed inside the company after the December 2025 model inflection, why nearly everyone at Shopify now uses AI tools daily, and what happens when code generation outpaces the rest of the software delivery pipeline.
We go deep on Shopify’s internal AI stack: Tangle for reproducible ML and data workflows, Tangent for auto-research loops that keep iterating toward better results, SimGym for customer simulation and theme optimization, and Liquid AI for ultra-low-latency and long-context workloads. Mikhail also reflects on PR review in the age of coding agents, the future of CI/CD, Sidekick Pulse, UCP, hiring at Shopify, and the untold story behind Sydney at Microsoft.
We discuss:
• Why Shopify’s AI usage effectively reached company-wide adoption
• The December 2025 inflection when model quality suddenly changed behavior
• Why token budgets matter, but only if paired with strong critique and review loops
• The real bottleneck in AI coding: PR review, CI/CD, and deployment stability
• Why more AI-written code can still mean more bugs in production
• Tangle: Shopify’s reproducible workflow system for ML and data experimentation
• Tangent: auto-research loops for optimizing pipelines, prompts, throughput, and infra
• Why auto-research is democratizing ML experimentation beyond researchers
• SimGym: simulating customers using Shopify’s historical data and browser-based agents
• Why customer simulation is hard to copy without years of real merchant behavior
• HSTU, counterfactuals, and modeling merchants and buyers over time
• UCP and Shopify’s product catalog as runtime infrastructure for agents
• Why Shopify is using Liquid AI for low-latency search and long-context distillation
• Sidekick Pulse, recsys, and where Shopify’s AI platform is heading
• Hiring areas at Shopify across ML, data science, and distributed systems
• The backstory of Sydney, personality shaping, and the early Bing chatbot era
—
Mikhail Parakhin
• LinkedIn: linkedin.com/in/mikhail-parakhin
• X: https://x.com/MParakhin
Timestamps
00:00:00 Introduction
00:02:05 Mikhail Parakhin, Shopify, and Microsoft
00:03:20 Shopify’s AI Adoption and the December 2025 Inflection
00:10:07 Why Token Count Alone Is a Bad Metric
00:12:59 Why Shopify Built Its Own AI PR Review System
00:16:16 Why Git, PRs, and CI/CD May Break in the Agent Era
00:20:29 Tangle: Shopify’s System for Reproducible ML Workflows
00:28:19 Tangent: Auto Research for Optimization and Experimentation
00:34:58 The Limits of Auto Research
00:38:40 Why Tangle, Tangent, and SimGym Compound Together
00:39:24 SimGym: Simulating Customers with Shopify’s Historical Data
00:44:51 The Infra Behind SimGym
00:48:04 Why Shopify’s Data Gives SimGym a Moat
00:49:34 Counterfactuals, HSTU, and Modeling Merchant Trajectories
00:55:34 UCP, Shopify Catalog, and Identity Linking
00:57:11 Liquid AI: Why Shopify Uses Non-Transformer Models
01:01:17 Real Shopify Use Cases for Liquid
01:05:04 Can Liquid Scale Into a Frontier Model?
01:06:52 Sidekick Pulse, Recsys, and Shopify’s AI Stack
01:09:29 Hiring at Shopify: ML, Data Science, and Databases
01:10:43 Sydney at Bing: Personality Shaping and AI Character
01:13:32 Closing Thoughts
![[State of AI Papers 2025] Fixing Research with Social Signals, OCR & Implementation — Team AlphaXiv](https://i.ytimg.com/vi/T0mZJjl_dsA/mqdefault.jpg "[State of AI Papers 2025] Fixing Research with Social Signals, OCR & Implementation — Team AlphaXiv
From late-night dorm-room hacking sessions at Stanford to building the most-used platform for navigating AI research, the founders of *AlphaXiv* have spent the last few years watching the archive firehose explode from thousands to 30,000+ papers per month—and building the intelligent layer on top that researchers, applied engineers, and even VCs now rely on to make sense of it all. We caught up with *Raj, Rayhahn, and the AlphaXiv team* live at *NeurIPS 2025* to dig into the origin story (spoiler: it started as a class project with a view comment button next to archive paragraphs), why they beat Hugging Face Papers by obsessing over UI and getting Lauras authors to comment directly on their own work, how theyve evolved from comments to feeds to benchmarks to AI assistants that answer paper questions using Geminis 1M context window, whats broken in the academic review process (20% AI-generated reviews at ICLR, quality collapsing under exponential submission growth), why *papers are becoming less important* than implementations, Docker containers, and real-world usability, their vision for making it trivial to spin up any trending papers code directly in your browser, and the power law they see every day: archive has 2.4M papers, but applied researchers only care about the top 0.1% that are actually implementable—and AlphaXiv is building the ranking, tooling, and social signals to surface exactly those.
We discuss:
* The *origin story:* from a late-night web dev class project (view comment buttons on archive PDFs) to going viral on LinkedIn and becoming a full-time startup backed by Stanford advisors
* Why AlphaXiv beat *Hugging Face Papers:* better UI, direct commenting on the paper itself, and early traction with high-signal authors like the Laura/DPO/Llama teams
* The evolution: *comments → feed → benchmarks → AI assistant,* using social signal (views, comments, Twitter) to filter the archive firehose and surface what actually matters
* How they handle *PDF parsing and OCR:* DeepSeek Coder for cost/accuracy, Mistral OCR APIs, and passing diagrams directly into multimodal models like Claude
* Why *semantic search alone fails:* 3M papers, buzzword overload, and the need to weight by social signal (views, engagement) to return actually relevant results
* The *state of academic publishing:* exponential submission growth (30k/month in CS), 20% AI-generated reviews at ICLR, review quality collapsing, and the rise of AI-written papers flooding the system
* Why *papers are becoming obsolete:* the future is implementations, Docker containers, and interactive sandboxes—not PDFs describing what code does
* The *power law:* 2.4M papers on archive, but applied researchers (Spotify, Expedia, Nintendo) only care about the top 0.1% that are actually implementable and useful
* AlphaXivs roadmap: *Docker containers for papers,* agent-automated setup, ranking by implementation ease, and making it trivial to spin up trending work directly in your browser
* Papers of the year: *Tiny Recursive Models (TRMs),* evolutionary strategies at hyperscale, Agent R1 (RL-trained agents for complex reasoning), and the rise of AI for science (Agent Laboratory, Bionemo)
* The *Qwen vs. DeepSeek* dynamic: Qwens breakout year, DeepSeeks continued influence (OCR models triggering a flurry of competitive releases), and the open-source fine-tuning ecosystem around both
* Why *AI reviewers* could help (Stanfords Agent Reviewer, using AI as a linter for clarity and quality) but assessing novelty and lit review remains hard without search + social signal
* The vision: AlphaXiv as the *tool layer for research*—not a social network, but the best place to discover, understand, and implement the ideas that actually matter in AI
—
AlphaXiv Team
* AlphaXiv: https://alphaxiv.org
* X: https://x.com/alphaxiv
Where to find Latent Space
* X: https://x.com/latentspacepod
* Substack: https://www.latent.space/
00:00:00 Introduction: The Origin Story of AlphaXiv
00:03:12 Beating Hugging Face Papers: What Made AlphaXiv Different
00:04:17 From Comments to Discovery: Building the Research Feed
00:04:55 PDF Parsing and OCR: The Technical Infrastructure
00:06:01 Becoming a Company: Beyond Papers to Research Artifacts
00:06:37 Benchmarks and State of the Art: Replacing Papers with Code
00:07:54 Personalization and Search: Building the Best Research Assistant
00:08:54 Docker Containers and Reproducibility: Making Papers Runnable
00:10:52 Favorite Papers: Tiny Recursive Models and Evolutionary Strategies
00:13:15 AI for Science: Agent Laboratory and the Future of Research Automation
00:18:41 Agent R1 and RL for Long-Horizon Tasks
00:20:54 Qwens Year and the Open Source Landscape
00:22:43 The Crisis in Academic Publishing: AI Slop and Review Quality
00:25:04 Search and Social Signals: Why Semantic Search Isnt Enough
00:28:15 The Future of Research: Beyond the PDF
00:31:17 Implementation Ease as the New Ranking Signal")
