"How Roy Resh Scaled Retail AI by Moving from Custom Pipelines to Configurable Computer Vision Systems"

“How Alexander Smirnoff Built Practical Enterprise AI Systems by Combining GenAI with Traditional NLP”

Episode Summary: In this episode of Engineering Choices You Have to Defend, host Nicola Onassis sits down with Alex Smirnoff to explore how enterprise AI systems can deliver real business value without replacing the proven infrastructure that already works. At Luminoso, Alex has spent years building large-scale NLP and text analytics systems that help enterprises analyze customer reviews, semantic search data, and large document collections. When generative AI rapidly entered the market, the company faced pressure from customers and stakeholders to “AI everything” overnight. Instead of rebuilding the platform around large language models, Luminoso chose a hybrid architecture that combined traditional NLP algorithms, semantic search, classification systems, and retrieval pipelines with modern GenAI reasoning capabilities. Alex explains why many older NLP tools still outperform LLMs for specific tasks like classification and keyword extraction, and how GenAI works best as an intelligent reasoning layer on top of existing systems. The conversation also explores hallucinations in enterprise environments, RAG pipeline design, grounding responses in source data, and the growing gap between flashy AI demos and production-ready enterprise systems. For engineering leaders, this episode highlights an important lesson: practical AI systems are rarely built by replacing everything — they succeed by combining proven infrastructure with new reasoning capabilities in thoughtful, cost-effective ways. Key Takeaways: * Traditional NLP tools still outperform LLMs for many specialized tasks * GenAI works best as a reasoning layer on top of existing systems * Hybrid AI architectures reduce cost and improve scalability * Enterprise AI systems must ground responses in customer data * RAG pipelines require careful tuning and retrieval quality management * Hallucination control is critical in business environments * There is a major gap between AI demos and production systems * Replacing entire platforms with GenAI often creates unnecessary complexity * Engineering teams should focus on business use cases, not AI hype * Successful AI adoption requires experienced implementation and planning Connect with Alex Smirnoff: * LinkedIn: Alex Smirnoff — linkedin.com/in/alex-smirnoff-34a13135 [http://linkedin.com/in/alex-smirnoff-34a13135] Listen Now & Subscribe: Apple Podcasts, Spotify, Amazon Music, YouTube, iHeartRadio, Captivate, or wherever you get your podcasts. "Engineering Choices You Have to Defend explores the real technical decisions behind AI systems, enterprise architecture, and scalable software engineering.

“How David Phipps Built AI-Powered Retail Systems by Prioritizing UX Over Feature Factories”

Episode Summary: In this episode of Engineering Choices You Have to Defend, host Nicola Onassis sits down with David Phipps to explore how engineering teams can scale AI-powered retail systems without sacrificing usability, reliability, or operational simplicity. Before joining Generation Tux, David helped build AI-driven digital signage and audience analytics systems that combined computer vision, edge computing, and point-of-sale integrations to measure customer engagement and advertising performance in physical retail environments. David shares how the company faced a critical decision after years of accumulating feature requests that made the platform increasingly difficult to use. Instead of continuing to add more features, the team committed to a complete UX and platform overhaul focused on simplicity, scalability, and fleet management. The conversation explores why usability became a competitive advantage, how Linux and Docker improved reliability at scale, and why AI-assisted development increases the importance of planning, architecture, and stakeholder alignment. For engineering leaders, this episode highlights an important lesson: the most valuable engineering decisions are often the ones that reduce complexity instead of adding to it. Key Takeaways: * UX and simplicity can outperform feature-heavy competitors * AI systems operating at the edge require reliability and low operational overhead * Feature factories often create long-term scalability problems * Managing large fleets requires strong architecture and automation * Stakeholder alignment is critical during platform redesigns * AI-assisted development increases the importance of planning and oversight * Simplifying workflows often creates more value than adding new features Connect with David Phipps: * LinkedIn: David Phipps — linkedin.com/in/dphipps [http://linkedin.com/in/dphipps] Listen Now & Subscribe: Apple Podcasts, Spotify, Amazon Music, YouTube, iHeartRadio, Captivate, or wherever you get your podcasts. "Engineering Choices You Have to Defend explores the real technical decisions behind AI systems, platform architecture, scalability, and engineering leadership.

"How Matt Lievertz Built Privacy-First AI Coaching Systems by Treating Compliance as a Core Product Strategy"

EPISODE SUMMARY: In this episode of Engineering Choices You Have to Defend, host Nicola Onassis sits down with Matt Lievertz, VP of Engineering at Cloverleaf, to explore how engineering teams can build AI-powered products that balance personalization, privacy, and enterprise trust. Cloverleaf combines behavioral assessments, workplace communication data, and AI-driven insights to help teams improve collaboration and performance. But handling personality data, coaching interactions, and workplace integrations introduced major technical and ethical challenges around privacy, compliance, and system design. Matt shares how a difficult enterprise compliance conversation in 2022 became a turning point for the company. Instead of treating privacy as a legal checkbox, Cloverleaf chose to build privacy protections directly into the architecture of the platform. That decision later positioned the company ahead of emerging regulations like GDPR, CCPA, and the EU AI Act. The conversation also explores how AI systems increase the complexity of privacy engineering, why minimizing personally identifiable information is becoming critical for enterprise AI adoption, and how simplifying platform architecture unlocked both scalability and partner growth. For engineering leaders, this episode highlights an important lesson: privacy and trust are no longer compliance features — they are foundational product decisions that directly impact scalability, enterprise adoption, and long-term platform resilience. KEY TAKEAWAYS: * Privacy becomes significantly more complex in AI-powered products * Enterprise trust requires going beyond minimum compliance standards * Building privacy into platform architecture reduces future regulatory risk * AI systems increase pressure around PII handling and data minimization * Treating compliance separately from engineering creates long-term risk * Simplifying platform architecture reduces regression risk and operational complexity * Unified systems scale more effectively than fragmented configuration models * Privacy-first design can become a competitive advantage in enterprise sales * Strong platform foundations reduce future engineering fire drills * AI trust depends on structure, filters, tokenization, and human oversight CONNECT WITH MATT LIEVERTZ: * LinkedIn: Matt Lievertz — linkedin.com/in/lievertz [linkedin.com/in/lievertz] * Website: Cloverleaf — cloverleaf.me [cloverleaf.me] LISTEN NOW & SUBSCRIBE: Apple Podcasts, Spotify, Amazon Music, YouTube, iHeartRadio, Captivate, or wherever you get your podcasts. "Engineering Choices You Have to Defend explores the real technical decisions behind regulated software, compliance, and AI integration, helping leaders build secure, auditable, and user-friendly systems."

"How Lavanya Elangovan Reduced Technical Debt by Embedding Security, Compliance, and Infrastructure Upgrades into Healthcare Engineering Workflows"

EPISODE SUMMARY: In this episode of Engineering Choices You Have to Defend, host Nicola Onassis sits down with Lavanya Elangovan to discuss the hidden engineering decisions required to maintain secure, compliant, and scalable healthcare platforms. Lavanya shares how a planned MongoDB upgrade quickly evolved into a full-stack modernization effort involving Ruby on Rails, infrastructure dependencies, and more than 40 libraries. Driven by both security certification requirements and product scalability goals, the project exposed the risks of accumulated technical debt in regulated environments. The conversation explores how her team approached the migration through phased rollouts, automated testing, security validation, and incremental infrastructure improvements built directly into the product roadmap. Lavanya also explains why AI-assisted development increases the importance of engineering rigor, human oversight, and deployment discipline. For engineering leaders, this episode highlights a critical lesson: technical debt is not just a maintenance issue; it directly impacts security, compliance, deployment confidence, and long-term business velocity. KEY TAKEAWAYS: * Healthcare engineering requires stronger compliance and security practices * Infrastructure upgrades often reveal hidden dependency risks * Technical debt slows deployment speed and reduces release confidence * Incremental modernization is safer than large “big bang” migrations * AI-assisted coding still requires strong human oversight and testing * Embedding infrastructure work into product roadmaps improves long-term scalability * Deployment confidence is a key indicator of platform health CONNECT WITH LAVANYA ELANGOVAN: * LinkedIn: Lavanya Elangovan — linkedin.com/in/lavanya-elangovan [linkedin.com/in/lavanya-elangovan] LISTEN NOW & SUBSCRIBE: Apple Podcasts, Spotify, Amazon Music, YouTube, iHeartRadio, Captivate, or wherever you get your podcasts. "Engineering Choices You Have to Defend explores the real technical decisions behind regulated software, compliance, and AI integration, helping leaders build secure, auditable, and user-friendly systems."

"How Roy Resh Scaled Retail AI by Moving from Custom Pipelines to Configurable Computer Vision Systems"

Episode Summary: In this episode of Engineering Choices You Have to Defend, host Nicola Onassis sits down with Roy Resh, VP of Engineering at https://traxretail.com?utm_source=chatgpt.comTrax Retail [https://traxretail.com?utm_source=chatgpt.com], to explore a pivotal architectural decision that reshaped how large-scale computer vision systems are built and scaled in retail environments. At Trax, Roy and his team built a computer vision platform that analyzes shelf images captured in retail stores, identifying products, pricing, and point-of-sale materials to generate a digital representation of store shelves. This enables brands to measure execution, shelf share, and product availability in near real time. But as the platform scaled across enterprise clients, complexity began to compound rapidly. What started as a unified recognition pipeline evolved into a heavily customized system, with per-client logic for attributes like expiration dates, display detection, reporting formats, and KPI calculations. Each new customer introduced new requirements, leading to custom code per client, duplicated processing flows, and increasingly long onboarding cycles that stretched from weeks to months. Roy explains how the system eventually reached a breaking point: onboarding delays of 30–60 days, rising operational overhead, and microservices becoming entangled with client-specific logic. In some cases, the platform even processed the same image multiple times to satisfy different customer requirements, driving up cost and complexity. The team made a strategic decision to move away from custom implementations and toward a configurable, JSON-driven workflow architecture. Built on event-driven microservices, queues, and coordination barriers, this new system allowed engineering teams to define and version entire processing flows through configuration rather than code. This shift enabled safer deployments, faster experimentation, and gradual rollouts per client—without affecting the entire platform. It also introduced a standardized KPI layer, reducing the need for bespoke reporting logic across customers. Roy also discusses the importance of human-in-the-loop validation in production AI systems. In a constantly evolving retail environment, human annotators help generate training data, validate model outputs, and maintain accuracy for high-stakes enterprise use cases where precision is critical. For engineering leaders, this episode highlights a key lesson: when every customer forces new code paths, you’re not scaling a product—you’re scaling complexity. Key Takeaways: * Over-customization is a clear signal of architectural scaling limits * Long onboarding cycles often reveal hidden system fragmentation * Configurable workflows reduce dependency on per-client code changes * Event-driven, JSON-based orchestration improves flexibility and deployment safety * Gradual migration strategies reduce risk in enterprise system rewrites * Standardizing KPI logic is as important as standardizing AI pipelines * Human-in-the-loop systems remain essential in dynamic real-world AI environments * Scalable platforms reduce variability instead of multiplying it Connect with Roy Resh: * LinkedIn: Roy Resh: linkedin.com/in/roy-resh [https://www.linkedin.com/in/roy-resh/] Listen Now & Subscribe: Apple Podcasts, Spotify, Amazon Music, or wherever you get your podcasts. "Engineering Choices You Have to Defend explores the real technical decisions behind regulated software, compliance, and AI integration, helping leaders build secure, auditable, and user-friendly systems."