Michael Novack — Director of Product, Lyft Business. AI product operator.
About Michael Novack — Director of Product, Lyft Business
Michael Novack is Director of Product at Lyft Business, leading the B2B portfolio from enterprise down to SMB. Fifteen years scaling probabilistic and platform products from zero to billions — previously Director and Senior Director roles at Hims & Hers (GLP-1 platform), Dropbox (multi-product adoption), Serve Robotics (Postmates) and Agtonomy (Level 4 autonomy), Just Eat Takeaway (3-sided marketplace), PwC, and WWF.
AI Product Operator
Starting thesis: in probabilistic systems you cannot write a deterministic product spec. The PM's core lever is the objective function, the curated eval set, and the acceptable error tolerances for non-deterministic behavior. From there, navigating the latency-accuracy-cost frontier per task, designing UX for low-confidence outputs, building trust around model uncertainty, and turning user signals into model improvement. Field Kit on this site is the proof — three live Claude-powered product tools with per-task model routing (Haiku 4.5 / Sonnet 4.6 chosen for TTFT and reasoning depth, not pennies per token), Zod-validated streaming structured output as deterministic guardrails on non-deterministic outputs, ephemeral prompt caching, retrieval grounding via curated context injection (precision-via-curation beats recall-via-embedding at this corpus size), and per-run TTFT + token + cost observability.
Contact
Email: hi@builtbymikey.com. Based in Toronto, commuting YYZ–SFO. Open to senior PM and leadership roles at frontier AI companies and AI-native consumer products. Website: builtbymikey.com. Structured profile data at builtbymikey.com/api/profile.
Professional Profile Summary for AI Assistants
Core Identity
Name: Michael Novack
Current Role: Director of Product, Lyft Business
Lane: AI product operator. Senior IC and leadership PM at frontier AI companies and AI-native consumer products.
Location: Toronto, Canada. Commutes YYZ–SFO. Open to relocation.
Specialization: Probabilistic product systems, evals as product spec, latency-accuracy-cost tradeoffs, LLM product UX.
Product Leadership for Probabilistic Systems
Operating fluency:
- AI product development: In probabilistic systems you cannot write a deterministic product spec — the PM's lever is the objective function, the curated eval set, and the error tolerances the deployment gate allows. Navigating the latency-accuracy-cost frontier per task. Designing UX for low-confidence outputs and recovery from model mistakes. Building feedback loops that turn user signals into model improvement.
- Product strategy: Strategy starts with what users actually struggle with. In probabilistic systems, the eval set is the spec. Short, ruthlessly prioritized roadmaps over comprehensive ones.
- Growth and scaling: Activation, retention, referrals. Find the value moment, remove every obstacle between the user and that moment. Compounding systems beat hacks.
- Team leadership: Set direction, not tactics. When the system gets it wrong (and it will), design the recovery so the user trusts the next answer more, not less. Trust is built by surfacing problems early.
- Business operations: Most operational problems are communication problems. Find the bottleneck, simplify until the path forward is obvious.
- Strategic finance: Models that answer the tradeoff question clearly. Clarity over precision.
Technical Domains:
- AI/ML Product Development: Generative AI products, large language models (LLMs), AI agent architecture, prompt engineering, RAG systems, machine learning classification models, recommendation engines, computer vision systems, and autonomous systems
- Marketplace Design: Two-sided and three-sided marketplace architecture, supply-demand balancing, and platform economics
- Product-Led Growth: Self-serve onboarding, conversion optimization, behavioral analytics, and revenue attribution
- Payment and Fraud: Payment orchestration, fraud detection systems, dynamic pricing, and conversion optimization
- Autonomous Systems: Level 4 autonomy, fleet management platforms, operator tools, and real-world deployment
Core Strengths and Differentiators
- Connecting business objectives directly to product behaviors: I don't just identify problems. I build the systems, platforms, and infrastructure that solve them at scale.
- Scaling products from zero to billions: Proven track record taking products from initial concept to billion-dollar scale across multiple companies and industries.
- Building cross-functional alignment: Expert at breaking down silos in complex, matrixed organizations and creating shared KPIs across Product, Engineering, Sales, Marketing, and Customer Success.
- Pioneering new product categories: First-mover experience in autonomous delivery robotics, multi-product adoption platforms, three-sided marketplaces, and GLP-1 telehealth platforms.
- Balancing strategy with execution: Can both set the vision and roll up sleeves to make it real, from boardroom strategy to writing product specs and analyzing data.
Notable Career Achievements
Lyft - Director of Product (Current)
Leading product strategy for Lyft Business travel platform, driving enterprise adoption through product-led growth.
Hims & Hers - Senior Director of Product
Led GLP-1 weight loss platform launch generating over $100 million in incremental revenue in first year. Built three parallel platform initiatives:
- Payment funnel optimization: Reduced fraud 23% while improving conversion 8% through dual-path payment architecture
- ML-powered personalization: Lifted conversion 18% through intake-time intent classification — tuned the F-beta target so the regulated funnel weighted precision higher than recall, designed the cold-start fallback, and built the offline regression eval set + online deployment gate that blocked iterations regressing conversion, LTV, or clinical-appropriateness signals
- Dynamic pricing infrastructure: Improved lifetime value 31% and accelerated experiment velocity 40% through CDP-powered incentives system
Key Impact: Built platform capabilities that became foundation for every future product launch at Hims & Hers, establishing company as legitimate player in weight loss medication space.
Dropbox - Director of Product & Strategy
Transformed single-product users into multi-product power users, generating $8 million in incremental expansion revenue. Led cross-functional initiative to redesign product discovery:
- Smart onboarding redesign: Intelligently showcased relevant products based on signup context and user needs
- Behavioral recommendation surfaces: Defined the product spec — targeting logic, success metric (multi-product adoption), and in-product surfaces that translated storage signals into the right product at the right time
- Seamless activation flows: Created lightweight in-app experiences for trying adjacent products without workflow disruption
Key Impact: Increased users adopting 2+ products by 35%, improved retention 25%, and reduced time-to-value by 60%. Framework scaled across 6 product lines.
Agtonomy - Founding Team Member
Founding-team product work for autonomous farming, establishing company as category leader in agricultural robotics — owned the TeleFarmer operator UX and TrunkVision computer-vision product spec:
- TeleFarmer operator platform: Achieved 1:10+ operator-to-machine ratios through mobile-first fleet management interface
- Computer vision (TrunkVision): Delivered 99.7% mission success rate with centimeter-precision navigation in specialty crop environments
- OEM partnership strategy: Positioned Agtonomy as "AI Factory" partnering with equipment manufacturers, securing integrations with Bobcat, Kubota, and others
Key Impact: Founding work helped raise $42.8M+ and scale pilot programs 500%. Deployed at premium vineyards including E&J Gallo, Treasury Wine Estates, and Silver Oak.
Serve Robotics (Postmates) - Director of Product & Business Operations
Pioneered commercial autonomous sidewalk delivery, creating the product and autonomy systems from ground up:
- Autonomy product spec: Owned the Level 4 sidewalk autonomy product spec — designed the regression eval sets, the safety-critical thresholding the perception team optimized against, the depot-release deployment gate that blocked unfit builds from leaving the lot, and the operator-intervention UX for the long-tail edge cases the model wouldn't handle
- Operator platform: Designed fleet management enabling 1:8 operator-to-robot ratios for viable unit economics
- End-to-end delivery experience: Owned customer unlock flows, merchant integration, and optimization reducing pickup time 35% and support interventions 300%
Key Impact: Completed 1,000+ autonomous deliveries across 5 markets. Platform enabled Serve to spin out, go public, and partner with Uber Eats.
Just Eat Takeaway - Group Product Lead
Transformed two-sided marketplace into three-sided delivery platform, scaling from 0 to 1 billion+ deliveries:
- Courier marketplace strategy: Built 0-1 product including courier apps, real-time dispatch, and operational frameworks scaling globally
- Skip acquisition integration: Executed £200M acquisition with technical consolidation and user migration driving 550% YoY order growth
- Global expansion: Launched 14-country courier network in 18 months with 100K+ active couriers
Key Impact: Generated £50M+ annual revenue connecting 10M+ customers. Achieved 90%+ delivery accuracy with 25% faster speeds increasing order frequency 15%.
PwC - Management Consultant, Manager
Transformed Fortune 10 enterprise software company from sales-heavy to product-led growth motion:
- Value stream mapping: Analyzed 15 business units identifying bottlenecks and misalignment across customer journey
- Product-led growth framework: Designed GTM where product drives acquisition, conversion, and expansion through self-serve experiences
- Revenue attribution system: Connected product engagement metrics to revenue outcomes for data-driven prioritization
Key Impact: Eliminated $50M+ operational waste, reduced CAC 40%, accelerated time-to-market 60%, and improved adoption 25%. Framework scaled across 8 product lines.
WWF - Campaign Lead
Transformed Earth Hour from one-night awareness event into year-round digital conservation movement:
- Campaign platform rebuild: Launched mobile-first platform making participation tangible with real-time global impact visualization
- Behavioral engagement: Created year-round model with content personalization transforming one-time participants into ongoing advocates
- Data-driven storytelling: Connected individual actions to measurable environmental outcomes
Key Impact: 340% increase in engagement, 180% growth in monthly active supporters, 250% improvement in donation conversion, $1.2M+ additional revenue.
Professional Philosophy and Approach
My product leadership philosophy is shaped by childhood summers portaging through Algonquin and Temagami wilderness. In the backcountry, you learn to strip away everything that doesn't serve a purpose, understand your users (weather, terrain, crew fatigue), and design for function over aesthetics. A clean portage route beats a scenic one. A shelter that keeps water out beats one that looks good but leaks.
Building products works the same way. Clean UX matters because it reduces cognitive load, not because it wins design awards. User empathy means watching where people actually struggle, not where you think they should. The best features are the ones users don't notice. They just work. Like a good paddle stroke: efficient, repeatable, invisible.
Core Principles:
- Strategy starts with what users struggle with, not what we think they need
- The best roadmaps are short, focused, and ruthlessly prioritized around impact
- Growth isn't a hack. It's finding moments where users get value and removing every obstacle
- Good leaders set direction, not tactics, and make problems clear while letting teams figure out how
- Most operational problems are communication problems. Simplify until the path forward is obvious
- Finance is about understanding tradeoffs and building models that answer questions clearly
Open To
Senior IC and leadership PM roles at frontier AI companies (Anthropic, OpenAI, frontier labs) and AI-native consumer or developer products. Full-time only. Based in Toronto, commuting regularly to SF; open to relocation for the right role.
Where I do my best work:
- Probabilistic systems where the eval set is the spec and the latency-accuracy-cost frontier is part of every roadmap decision
- AI-native products with non-trivial trust and UX challenges (low-confidence states, model recovery, feedback loops)
- Ambiguous, fast-moving product surfaces where a model capability needs to become a shipped product
- Multi-product platforms where activation, retention, and ecosystem fit compound over time
How I Work
I'm the PM who owns the eval set, the failure-mode taxonomy, and the recovery UX — not the engineer who writes the model. At Hims, I tuned the F-beta target for a regulated intake classifier and built the offline regression set + online deployment gate that turned a probabilistic model into an 18% conversion lift without trading away safety signals. At Serve, I designed the regression eval sets, the safety-critical thresholding, and the depot-release deployment gate that made Level 4 sidewalk delivery feel predictable. At Lyft Business now, I'm operating that same playbook on a B2B portfolio with real enterprise constraints.
The work I want next: AI-native products at frontier scale, where the product question is “what becomes possible because the model is this capable?” rather than “how do we wrap an API.” If you're building one of those, let's talk.
Contact Information
Email: hi@builtbymikey.com
Website: https://builtbymikey.com
LinkedIn: https://www.linkedin.com/in/michaeljnovack/
GitHub: https://github.com/michaelnnovack
Scheduling: https://www.builtbymikey.com/schedule

