Token Engineering

NeuroMetric AI
  • 5.0 (1)
  • Technology
  • Updated Daily

A show about AI economics and the cost of thinking. From Neurometric, on the tokenminning movement, the discipline emerging to make every unit of AI work cheaper, faster, and more reliable.

  1. Jun 16

    Automating the SLM Development Loop: Inside the Pioneer Agent Paper with Yash Sharma of Neurometric AI

    In this episode of Inference Time Tactics, Cooper sits down with Yash Sharma, Head of AI Research at Neurometric AI, to break down the Pioneer Agent paper from Fastino Labs—a system that uses Claude Sonnet as an ML engineer in a box to build and improve small language models end-to-end. From cold start data curation to production failure diagnosis, the paper argues the real bottleneck in SLM creation isn't training—it's everything around it.   We talked about:   What Pioneer Agent actually is and why Fastino built a 25-page system paper instead of just a training paper.  How Neurometric is already tackling the same problems—and where the paper maps onto our approach.  Why naive retraining on production data quietly degrades your model—and how an agentic loop fixes it.  What the benchmarks reveal about where SLMs work out of the box versus where they need intervention.  Where Pioneer Agent hits its limits—and how Neurometric is pushing SLMs toward harder agentic tasks. Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social   Host/s: Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Guest/s: Yash Sharma https://x.com/yash_j_sharma  https://www.linkedin.com/in/yashjsharma

    32 min

  2. May 29

    SLMs Beat GPT-4o: How AI Is Being Used in Hiring with Fletcher Wimbush, CEO of Discovered AI

    In this episode of Inference Time Tactics, Cooper sits down with Fletcher Wimbush, CEO of Discovered AI, to get into what’s actually broken about hiring—and how 13 years of bootstrapped recruiting, 10,000 interviews, and four peer-reviewed journal articles became the foundation for a platform now outperforming its biggest competitors. With hundreds of applicants flooding every open role and general AI tools hitting the ceiling at 70-80% resume screening accuracy, Discovered is combining behavioral science with fine-tuned small language models to get hiring right the first time.   We talked about:   The hidden cost of over-relying on qualifications over attitude and integrity—and why Warren Buffett says a smart person with low integrity is your biggest threat. How 10,000 interviews and four peer-reviewed journal articles with Bowling Green State University gave Discovered’s hiring system its scientific backbone. Fine-tuned small language models hitting 98.5% accuracy on resume screening where GPT-4o topped out at 70-80%—at a fraction of the cost. What’s coming next on the Discovered roadmap: smarter resume database search, AI-powered parsing, and an AI interviewer built to ask the right follow-up questions.   Connect with Discovered AI: Website: https://discovered.ai  LinkedIn: https://www.linkedin.com/in/fletcher-wimbush  Email: fletcher@discovered.ai   Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social   Hosts: Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc

    28 min

  3. Mar 9

    Voice Intelligence at Scale: From Call of Duty to Fraud Detection with Modulate AI

    Every day billions of voice conversations happen across games, customer service calls, and financial transactions. Almost none of them are understood by machines. In this episode of Inference Time Tactics, Calvin Cooper and Yash Sharma sit down with Carter Huffman, CTO and co founder of Modulate, to explore the AI systems that can finally understand voice conversations in real time.   Modulate’s model Velma 2.0 powers voice intelligence across industries. From moderating voice chat in games like Call of Duty to detecting fraud in financial calls and analyzing customer support conversations, their system uses ensembles of specialized models to capture tone, intent, emotion, and conversational dynamics. Instead of relying on giant foundation models, Velma orchestrates over 100 specialized models to deliver higher accuracy at dramatically lower cost.   We talked about:   The challenge of processing a trillion hours of annual global voice traffic. Scaling real-time moderation for massive platforms like Call of Duty. Capturing nuance, tone, and sarcasm beyond basic text transcripts. Ensemble architecture utilizing over 100 specialized models. Orchestration layers that trim compute costs by identifying optimal model subsets. Achieving order-of-magnitude cost savings compared to large foundational models. Applying "exploration vs. exploitation" optimization to shifting conversation data. Future development of "context graphs" to map participant intent and causality. Resources Mentioned: NeuroMetric Audio Leaderboard: https://leaderboard.neurometric.ai/?leaderboard=audio  Connect with Modulate: Website: https://www.modulate.ai/  LinkedIn: https://www.linkedin.com/in/carter-huffman-a9aba05b  Velma: https://www.modulate.ai/velma  Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social Hosts: Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Yash Sharma https://x.com/yash_j_sharma  https://www.linkedin.com/in/yashjsharma/

    32 min

  4. Jan 16

    From GPU Scarcity to GPU Waste: Solving the Utilization Crisis

    In this episode of Inference Time Tactics, Cooper and Byron sit down with Charlie and Anil from Rapt AI to tackle one of the industry's most expensive problems: GPU underutilization. With half a trillion dollars invested in GPU infrastructure running at just 20-30% utilization, Rapt AI is building AI-powered orchestration that automatically analyzes workloads and matches them to the right compute resources—no guesswork required.   We talked about:   Why half a trillion dollars in GPU infrastructure runs at only 20-30% utilization—and how a 5% drop costs $200,000 per $2M investment.  How Rapt AI's platform continuously analyzes workloads and auto-optimizes GPU allocation, letting customers run 4-14 models per GPU.  Real results: moving workloads from H100s to A100s at 40% of the cost, and reducing GPU footprints from 184 to under 50 while improving performance.  Why 2026 becomes the year of inference as agentic workloads create unprecedented infrastructure chaos.  The shift from supply problems to optimization problems—and why abstraction layers matter across multi-vendor environments.  Power as the next crisis: tokens-per-watt emerging as the critical metric alongside tokens-per-dollar.  How intelligent orchestration frees up data scientists and ML ops teams from infrastructure tuning to focus on AI innovation. Connect with Rapt AI: Website: https://www.rapt.ai/  LinkedIn (Anil Ravindranath): https://www.linkedin.com/in/anilravindranath  LinkedIn (Charlie Leeming): https://www.linkedin.com/in/charlieleeming/  Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social Hosts: Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Byron Galbraith https://x.com/bgalbraith  https://www.linkedin.com/in/byrongalbraith

    40 min

  5. 12/22/2025

    Lessons from the Leading Edge: What 420 AI Deployments Reveal About Enterprise Success

    In this episode of Inference Time Tactics, Rob, Cooper, and Byron sit down with Shawn Rogers, CEO of BARC US to unpack fresh data from 421 organizations actively deploying AI in production. Shawn shares what separates the 20% of AI leaders from everyone else, why cost surprises are hitting harder than expected, and how the pressure to "just do AI" is causing companies to skip critical foundations—often to their detriment. We talked about:   Why multi-model strategies and small language models are becoming essential for enterprise AI. The seven foundational areas that help AI leaders deploy twice as many projects as everyone else.  Why 51% of deployments face unexpected cost overruns—and which expenses hit hardest.  Data quality jumping to the #1 challenge, affecting 44% of production deployments.  The IT satisfaction paradox: top resource at the start, lowest satisfaction scores at scale.  How responsible AI priorities shifted as human-in-the-loop dropped from 36% to 21%.  Resources Mentioned: Lessons from the Leading Edge: Successful Delivery of AI/GenAI https://barc.com/research/successful-ai-genai-delivery/ Connect with BARC: Website: https://barc.com/  LinkedIn (Shawn Rogers): https://www.linkedin.com/in/shawnrogers/  Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social Hosts: Rob May https://x.com/robmay  https://www.linkedin.com/in/robmay   Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Byron Galbraith https://x.com/bgalbraith  https://www.linkedin.com/in/byrongalbraith

    44 min

  6. 12/16/2025

    The Thinking Algorithm Leaderboard: Why No Single Model Wins

    In this episode of Inference Time Tactics, Cooper and Byron break down NeuroMetric's Thinking Algorithm Leaderboard and what it reveals about building production-ready AI agents. They share why prompt engineering with a single model won't cut it for enterprise use cases, explore the impact of inference-time compute strategies, and discuss what they learned from testing 10 models across real CRM tasks—from surprising token inefficiency to catastrophic failures in SQL generation.   We talked about:   Why NeuroMetric built the first leaderboard combining models with inference-time compute strategies.  How Salesforce's CRMArena-Pro reflects real multi-step business tasks better than pure reasoning benchmarks.  The jagged frontier: no single model or technique dominates across all tasks.  Why GPT 20B was surprisingly token inefficient—twice as slow as GPT 120B for similar accuracy.  How GPT-5 nano's conversational style broke SQL generation tasks completely.  Trading accuracy for speed: two-model ensembles versus five, and saving 20+ seconds per task.  Throughput constraints as a hidden bottleneck when scaling to production volumes.  Future directions: LLM-guided search, task clustering, and compression to specialized small models. Resources Mentioned: CRMArena-Pro from Saleforce: https://www.salesforce.com/blog/crmarena-pro/ Thinking Algorithm Leaderboard:  https://leaderboard.neurometric.ai/  Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social   Hosts: Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Guest/s: Byron Galbraith https://x.com/bgalbraith  https://www.linkedin.com/in/byrongalbraith

    29 min

  7. 11/05/2025

    Benchmarking Generalization: How AI Learns Beyond Training Data

    In this episode of Inference Time Tactics, Rob and Cooper from Neurometric sit down with Yash Sharma, an AI researcher whose work is reshaping how we understand model generalization. Yash recently completed his PhD at the Max Planck Institute for Intelligent Systems and has held research roles at Google Brain, Meta AI, Amazon, Borealis AI, and IBM Research. His studies on compositional generalization, adversarial robustness, and long-tail benchmarks reveal when and why models succeed—or fail—at reasoning beyond their training data. If you’re designing inference-time systems, building agents that need reliability, or just want to understand what “generalization” actually means in practice, this conversation bridges deep theory with actionable insight—clear, technical, and strategically grounded. Key Topics What it really means for AI systems to generalize beyond their training data Why large language models still fail in novel or unpredictable scenarios How inference-time compute can both amplify and reveal generalization limits What these limits mean for building reliable, agentic AI systems How to benchmark generalization in real-world settings Yash’s “Let It Wag!” benchmark for testing long-tail and under-represented concepts Why genuine scientific breakthroughs (like curing cancer) require more than scaling test-time compute Connect with Yash Sharma: Yash Sharma Let It Wag! Benchmark Paper: Pretraining Frequency Predicts Compositional Generalization of CLIP (NeurIPS 2024 Workshop) Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social   Rob May https://x.com/robmay  https://www.linkedin.com/in/robmay   Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc

    37 min

  8. 10/03/2025

    Solving the Cold Start Problem in AI Inference

    In this episode of Inference Time Tactics, Rob, Cooper, and Byron sit down with Prashanth Velidandi, co-founder of InferX, to explore how serverless inference is tackling the AI “cold start problem.” They dig into why 90% of the model lifecycle happens at inference—not training—and how cold starts and idle GPUs are crippling efficiency. Prashanth explains InferX’s snapshot technology, what it takes to deliver sub-second cold starts, and why inference infrastructure—not just models—will define the next era of AI. We talked about:   Why inference represents 90% of the model lifecycle, compared to the training focus most of the industry has. How cold starts and idle GPUs create massive inefficiencies in AI infrastructure. InferX’s snapshot technology that enables sub-second model loading and higher GPU utilization. The challenges of explaining and selling deeply technical infrastructure to the market. Why enterprises care about inference efficiency, cost, and reliability more than model size. How serverless inference abstracts away infrastructure complexity for developers. The coming explosion of multi-agent systems and billions of specialized models. Why sustainable innovation in AI will come from inference infrastructure. Connect with InferX Prashanth Velidandi https://inferx.net  https://x.com/pmv_inferx  https://www.linkedin.com/in/prashanth-velidandi-98629b115 Connect with Neurometric: Website: https://www.neurometric.ai/  Substack: https://neurometric.substack.com/  X: https://x.com/neurometric/  Bluesky: https://bsky.app/profile/neurometric.bsky.social   Rob May https://x.com/robmay  https://www.linkedin.com/in/robmay   Calvin Cooper https://x.com/cooper_nyc_  https://www.linkedin.com/in/coopernyc   Byron Galbraith https://x.com/bgalbraith  https://www.linkedin.com/in/byrongalbraith

    35 min
See All (15)

About

A show about AI economics and the cost of thinking. From Neurometric, on the tokenminning movement, the discipline emerging to make every unit of AI work cheaper, faster, and more reliable.

Information

  • Creator
    NeuroMetric AI
  • Years Active
    2025 - 2026
  • Episodes
    15
  • Rating
    Clean
  • Copyright
    © Copyright 2025 All rights reserved.
  • Show Website
    Token Engineering

You Might Also Like