<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Gurinderpal Singh on Medium]]></title>
        <description><![CDATA[Stories by Gurinderpal Singh on Medium]]></description>
        <link>https://medium.com/@gurinder21?source=rss-6d90a56c2057------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*cL1njC8slT7Vjj7dK7zT1w.png</url>
            <title>Stories by Gurinderpal Singh on Medium</title>
            <link>https://medium.com/@gurinder21?source=rss-6d90a56c2057------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 22:40:42 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@gurinder21/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[The Vendor Change That Wasn’tSupposed to Matter]]></title>
            <link>https://medium.com/@gurinder21/the-vendor-change-that-wasntsupposed-to-matter-c01df4ed465c?source=rss-6d90a56c2057------2</link>
            <guid isPermaLink="false">https://medium.com/p/c01df4ed465c</guid>
            <category><![CDATA[supply-chain]]></category>
            <category><![CDATA[quality-assurance]]></category>
            <category><![CDATA[biotech]]></category>
            <category><![CDATA[validation]]></category>
            <category><![CDATA[manufacturing]]></category>
            <dc:creator><![CDATA[Gurinderpal Singh]]></dc:creator>
            <pubDate>Wed, 30 Sep 2026 20:16:18 GMT</pubDate>
            <atom:updated>2026-09-30T20:16:18.665Z</atom:updated>
            <content:encoded><![CDATA[<p>Single-use systems promised speed. But in GMP manufacturing, supplier changes can quietly move qualification risk outside your walls.</p><p><em>A practical look at how single-use technology changed GMP risk — not by eliminating qualification work, but by moving critical variables into the supplier network.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/1*eqixJNWnwlhk938ygd0GKg.png" /><figcaption><em>Single-use assemblies can accelerate GMP readiness, but their qualification history depends on changes made far beyond the manufacturing floor.</em></figcaption></figure><p>Somewhere, right now, a supplier is filing a routine change notice: a new resin lot, a tweaked extrusion process, a different gamma sterilization contractor. To them, it’s paperwork. To the manufacturer receiving it, it can be the difference between a clean batch record and a full-blown investigation.</p><p>FDA has cited exactly this failure mode. In one recent warning letter, a contract manufacturer released multiple device lots without documented justification for skipping revalidation, after quietly switching to a new contract sterilizer months earlier. The sterilization method hadn’t changed on paper. The provider had. Nobody treated that as the qualification event it actually was, until FDA did.</p><p>That’s the part of the single-use story nobody tells at conferences.</p><p><em>The equipment is disposable. The qualification burden is not.</em></p><h3>The Pitch Was Always Speed</h3><p>Single-use systems have been standard in biomanufacturing for the better part of a decade now. The pitch was simple: no more cleaning validation, no more shared-equipment cross-contamination risk, faster turnaround between batches, faster turnaround from empty room to GMP-ready. And on those terms, single-use delivered. I’ve built sites around exactly that promise.</p><p>But “single-use” was always a slightly misleading name. The equipment is disposable. The qualification burden is not. It didn’t disappear when we stopped cleaning stainless steel tanks. It moved, from something you controlled inside your own four walls, to something a supplier control inside theirs.</p><h3>Where the Risk Actually Went</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/1*rVFa96JysTnBN469HL93Kg.png" /><figcaption><em>In disposable systems, risk travels through resin suppliers, component fabricators, sterilization partners, and the final GMP site.</em></figcaption></figure><p>When you validate a stainless steel system, the variables are yours: your cleaning procedure, your equipment, your data. When you qualify a single-use system, a meaningful share of the variables belongs to someone else. The bag manufacturer’s resin supplier. The filter’s membrane lot. The gamma irradiation contractor’s dose mapping. Every one of those is a potential change point, and most of them happen without you in the room.</p><p>The industry built a mechanism for this: change notifications. In theory, a supplier flags any change that could affect form, fit, or function, and you assess whether it triggers revalidation. In practice, change notifications arrive in volume, often without enough technical detail to actually assess risk, and the assumption that “it’s the same part number” quietly substitutes for real evaluation. That’s precisely the gap in the warning letter I mentioned: a documented, defensible justification for not revalidating was never made. Someone just assumed it wasn’t necessary.</p><h3>This Is a Qualification Problem, Not a Procurement Problem</h3><p>The instinct is to treat vendor changes as a supply chain issue, something quality reviews after the fact. I’ve come to see it differently. A single-use component with an active change history is closer to a piece of equipment with an open deviation. It needs the same discipline: a defined risk assessment trigger, a traceability path back to the original qualification data, and a real technical evaluation before a lot gets used, not a rubber stamp because the part number matches.</p><p>That means building your qualification strategy with the assumption that the vendor will change something, eventually, quietly, and possibly without realizing it matters to you. Extractables and leachables profiles tied to specific resin formulations. Integrity testing that isn’t just a pass/fail checkbox but a comparison against the qualified baseline. A traceability matrix that actually connects a specific component lot to the validation data that justified using it, not just a generic reference to “the qualified single-use assembly.”</p><h3>The Trade Nobody Put in the Sales Deck</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/1*_WhjpCWYixuSPKwtlc7n3g.png" /><figcaption><em>Single-use systems replaced some internal controls with supplier-dependent controls — and that tradeoff needs to be managed deliberately.</em></figcaption></figure><p>None of this means single-use was the wrong call. It almost always still is the right call, for speed, for flexibility, for avoiding cross-contamination between products in a multi-product facility. But it was sold, implicitly, as qualification risk going down. What actually happened is qualification risk changed shape. It became distributed, quieter, and dependent on a supplier relationship instead of an internal procedure you fully control.</p><p>The sites that get this right build vendor change management into their qualification program from day one, not as a compliance afterthought, but as core scope, the same way they’d treat requalifying a piece of fixed equipment after a modification. The sites that get it wrong find out the hard way, usually when an auditor asks for the revalidation justification for a change that happened eighteen months ago, and the honest answer is that nobody looked closely enough to know it mattered.</p><p>Speed to GMP was never free. Single-use systems just moved where the bill comes due.</p><p><em>Source: FDA Warning Letter, </em><a href="https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/microspecialties-llc-707614-05082025"><em>fda.gov/…/microspecialties-llc-707614–05082025</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c01df4ed465c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Six Inches That Undid a Sterile Room]]></title>
            <link>https://medium.com/@gurinder21/the-six-inches-that-undid-a-sterile-room-5153fae57a69?source=rss-6d90a56c2057------2</link>
            <guid isPermaLink="false">https://medium.com/p/5153fae57a69</guid>
            <category><![CDATA[cleanroom]]></category>
            <category><![CDATA[sterile-processing]]></category>
            <category><![CDATA[design]]></category>
            <category><![CDATA[fda]]></category>
            <category><![CDATA[pharmaceutical]]></category>
            <dc:creator><![CDATA[Gurinderpal Singh]]></dc:creator>
            <pubDate>Fri, 21 Aug 2026 01:11:29 GMT</pubDate>
            <atom:updated>2026-08-21T01:11:29.367Z</atom:updated>
            <content:encoded><![CDATA[<p>A few months ago, FDA posted a warning letter that stuck with me. An inspector watched sterile vials get carried out of an ISO Class 5 hood, the most protected zone in the building, through an ISO Class 7 room on the way to further processing. For that stretch, the product was exposed to meaningfully worse air quality than its own classification required. The company’s name was redacted, but the finding wasn’t unusual. It was a facility with the right rooms, the right classifications, the right filters, undone by how material physically moved between them.</p><p>That’s the thing about clean rooms most people never see: the room passing inspection on paper and the room actually protecting the product, moment to moment, are two different achievements. Most of my career has been spent in the gap between them. So let me explain what a clean room actually is, the way I wish someone had explained it to me before I spent fifteen years living inside these rooms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/936/1*qv9e4o5Q0Mf5kudesitLGg.png" /><figcaption><strong>Figure 1:</strong> Typical Cleanroom Design</figcaption></figure><p><strong>It starts with a simple word: sterile</strong></p><p>In everyday language, “sterile” and “clean” mean roughly the same thing. Wipe the counter, sterile enough. In pharmaceutical manufacturing, they are not the same thing at all, and the gap between them is the entire reason clean rooms exist.</p><p>Clean means visibly free of dirt. Sterile means free of any living microorganism, at a probability so close to zero that regulators are comfortable calling it zero. A sterile product is one where the chance of a single organism surviving in it is vanishingly small, something like a one-in-a-million shot, sometimes tighter.</p><p>That distinction matters when the product is going into someone’s bloodstream, spinal fluid, or eye. A contaminated countertop is an inconvenience. A contaminated vial of an injectable medicine is a patient safety event, exactly what was on the line in that warning letter: not a paperwork gap, a real path for contamination to reach a sterile product.</p><p>We don’t leave “clean enough” to judgment, either. Rooms are classified against ISO 14644–1, mapped in pharma to letter grades, Grade A being the most critical zone where sterile product is actually open to the room, down through B, C, and D as risk drops. Each grade has a defined limit on how many particles of a given size are allowed per cubic meter of air. That number is the line the warning letter said had been crossed during a routine transfer.</p><p><strong>The room itself is quietly doing a lot of work</strong></p><p>A clean room isn’t clean because someone scrubs it constantly. It’s clean because of invisible systems working around the clock.</p><p>Air is the biggest one. It’s pulled in, filtered through HEPA material that traps particles far smaller than the eye can see, and pushed back out, again and again. A Grade A zone runs unidirectional airflow sweeping the space continuously; supporting rooms generally run 20 or more air changes per hour, climbing from there by classification. The room is, in effect, constantly flushing itself.</p><p>Then there’s pressure. Clean rooms run higher pressure than the hallway or an adjacent, less-clean room, on the order of 10 to 15 pascals between grades, and at least 12.5 pascals where a classified room borders unclassified space. Crack the door and air rushes out rather than in, carrying contamination away from the product. String several rooms together, each slightly higher pressure than the last, and you get a pressure cascade, an entire building quietly protecting its most critical point, the exact system undermined the moment product moves through a room it shouldn’t.</p><p>And finally, gowning. That gear isn’t for the comfort of the wearer. Humans are, from a contamination standpoint, disasters, shedding skin cells and microorganisms constantly without noticing. Gowning contains us, the messiest thing in the room, so what we shed stays inside the suit instead of settling onto the product.</p><p>None of this is trusted on faith. Rooms are continuously monitored for nonviable particles, the physical debris a counter tracks, and viable particles, living organisms picked up by air samplers and settle plates at defined, risk-based locations.</p><p><strong>Why the rules feel so unforgiving</strong></p><p>If you’ve wondered why pharmaceutical manufacturing seems so strict about things that look minor, a door propped open a beat too long, a glove touching a surface it shouldn’t, this is why. Small mistakes don’t stay small. A single transfer through the wrong room can introduce a contamination risk no amount of downstream testing can rule out.</p><p>That’s not paranoia. It’s proportional caution to the stakes involved. Design gets you qualified. Operational discipline, the kind that catches a transfer path nobody thought to question, is what keeps you that way.</p><p><strong>The part nobody sees</strong></p><p>What people don’t usually picture is the work required to prove all of this actually works before a single vial is filled. Qualification is where I’ve spent much of my career: smoke studies to visualize airflow, pressure cascade checks, HEPA filter integrity testing, and confirmation that a room performs as designed under real operating conditions. That’s the IQ, OQ, PQ sequence, and only after all three hold up does a room earn the right to make product for a patient. But qualification is not the finish line. Once in use, continuous environmental monitoring must show the room stays in control through nonviable particle monitoring and viable air, surface, and personnel sampling at risk-based locations and frequencies appropriate to the cleanroom grade. The program has to provide a similar level of assurance, showing the room continues to maintain the grade it was qualified for during actual operations.</p><p>It’s slow, exacting work, and it rarely gets talked about outside the industry. But it’s the reason you can trust that an injectable medicine, made in a room you’ll never see, by people you’ll never meet, is safe when it reaches you. That’s what I want to write more about, starting here: what a clean room actually is, and why an entire industry organizes itself around protecting something as important as clean air.</p><p><em>Source: FDA Warning Letter, </em><a href="https://www.fda.gov/media/192056/download"><em>fda.gov/media/192056/download</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=5153fae57a69" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Can Draft My Protocols. Here’s My Line Between Drafting and Deciding.]]></title>
            <link>https://medium.com/@gurinder21/ai-can-draft-my-protocols-heres-my-line-between-drafting-and-deciding-2ff67d0bc5cc?source=rss-6d90a56c2057------2</link>
            <guid isPermaLink="false">https://medium.com/p/2ff67d0bc5cc</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[biotechnology]]></category>
            <category><![CDATA[pharmaceutical]]></category>
            <category><![CDATA[validation]]></category>
            <category><![CDATA[quality-assurance]]></category>
            <dc:creator><![CDATA[Gurinderpal Singh]]></dc:creator>
            <pubDate>Tue, 11 Aug 2026 20:39:53 GMT</pubDate>
            <atom:updated>2026-08-11T20:39:53.392Z</atom:updated>
            <content:encoded><![CDATA[<p>Everyone’s talking about AI right now, and most of it has nothing to do with my world. I don’t spend my day thinking about chatbots. But it has quietly started showing up in commissioning, qualification, and validation (CQV) work, and I think it’s worth talking about honestly, because the conversation tends to swing between two extremes. Either AI is about to replace validation engineers, or it has no business anywhere near a regulated document. I don’t think either one is right.</p><p><strong>Where it actually helps</strong></p><p>A lot of C&amp;Q work isn’t complicated, it’s just tedious. Formatting Installation Qualification (IQ) and Operational Qualification (OQ) test scripts. Turning an engineer’s rough commissioning notes into proper protocol language, with acceptance criteria written the way a quality document needs them written. Rewriting the same boilerplate section for the tenth time with small edits. None of that is where my judgment adds value, but it eats a lot of time.</p><p>That’s where something like Copilot earns its keep. Give it a rough set of test steps and it’ll structure them into a protocol format, draft acceptance criteria language for review, clean up a table. It’s not writing the protocol. It’s clearing the administrative weight off the desk so the people who actually understand the system, and the risk behind it, can spend their time on the parts that matter.</p><p>That’s a real shift, and I don’t think it’s going away. But I’ve learned some lines I won’t cross, and I think they’re worth saying out loud.</p><p><strong>Your proprietary data doesn’t belong in a public tool</strong></p><p>Commissioning documents are full of information that’s specific to how a company actually builds and runs its facility, system specs, process parameters, layouts. That’s not something that should ever get pasted into a public AI tool sitting outside your company’s own environment. Before I let AI touch anything real, the first question is simple: is this an internal, approved system, or is it a public tool where I have no idea where that data goes. If it’s the second one, the answer is no, no matter how good the output would be.</p><p><strong>A draft can look complete and still be wrong</strong></p><p>Here’s where it gets more specific to this field. A qualification protocol isn’t just a document, it’s the output of a risk-based process. Under a framework like ASTM E2500, which is how most of the industry approaches commissioning and qualification now, you start with a user requirements specification, run a risk assessment to identify which requirements are critical to product quality or patient safety, and then build your test coverage around that risk assessment, not around a generic template. Every critical requirement has to trace to a test that actually verifies it. That traceability matrix, requirement to test to result, is what an auditor will pull first.</p><p>AI is genuinely good at producing a traceability matrix that looks complete. It’s much weaker at knowing which requirements are actually critical in the first place, because that judgment comes from the risk assessment, not from the document structure. I’ve seen a draft matrix where every listed requirement had a test mapped to it, cleanly, professionally formatted, and it still missed the point, because a requirement that should have come out of the risk assessment as critical had never been entered into the matrix at all. Nothing looked broken. The gap wasn’t a wrong answer, it was a question that was never asked, because the AI was working from the requirements it was given, not from an independent risk-based judgment about what those requirements should have included.</p><p>That’s a meaningfully different failure than a typo or a missing signature. It’s invisible on the page, and it’s exactly the kind of thing a reviewer who’s rushing, because the draft looks thorough, is most likely to miss.</p><p><strong>This isn’t hypothetical anymore</strong></p><p>Earlier this year, the FDA issued a warning letter to a drug manufacturer after an inspection found that process validation, one of the most basic requirements in this industry, had simply never been performed before product was distributed. When investigators asked why, the firm’s answer was that they’d used AI agents to help build their specifications, procedures, and records, and the AI had never flagged that validation was required. Nobody at the company had gone back and checked that against the actual regulation themselves.</p><p>The FDA’s response was blunt: an AI knowledge gap is not a defense. If you use AI to help create a document, you’re still required to review it and confirm it’s accurate and compliant, under your own quality system, and skipping that step is a violation on its own, separate from whatever else went wrong.</p><p>That’s the real danger with these tools, in a nutshell. A confident, well-structured draft is more dangerous than a rough one, because it’s easier to trust too quickly. It doesn’t tell you what it left out, and it won’t flag a gap in its own risk logic. The verification work doesn’t get shorter just because the draft got faster.</p><p><strong>Accountability doesn’t transfer to the tool</strong></p><p>“The AI didn’t tell us” isn’t an answer that holds up in an inspection, and it shouldn’t. Whoever signs a document, and whoever’s Quality Unit released it, owns what’s in it and what’s missing from it, full stop. That includes the data integrity trail around it, too. If a system is touching GxP records, ALCOA+ principles and audit trail expectations under 21 CFR Part 11 still apply to how that AI-assisted draft moved through review, not just to the final signature.</p><p>That’s the part I think gets lost in a lot of the excitement. AI can produce a solid starting draft, and it can build a matrix that looks airtight. It can’t run the risk assessment that should have shaped that matrix in the first place, and it can’t replace actually knowing your system, your procedures, and your product well enough to know what’s missing.</p><p><strong>Where I’ve landed</strong></p><p>I don’t think the answer is refusing to use these tools, and I don’t think it’s trusting them blindly either. It’s letting AI take the administrative weight off, while keeping the risk-based thinking, the traceability, and the verification exactly as rigorous as they’ve always had to be. The tools will keep getting better at sounding right. Our job is making sure sounding right and being right don’t quietly become the same thing in our heads.</p><p><em>Reference: </em><a href="https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/purolea-cosmetics-lab-722591-04022026"><em>FDA Warning Letter, April 2, 2026</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2ff67d0bc5cc" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[We Built the Manufacturing Capacity. We Forgot to Build the People.]]></title>
            <link>https://medium.com/@gurinder21/we-built-the-capacity-we-forgot-to-build-the-people-1f258dfd84dd?source=rss-6d90a56c2057------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f258dfd84dd</guid>
            <category><![CDATA[quality-assurance]]></category>
            <category><![CDATA[biotech]]></category>
            <category><![CDATA[manufacturing]]></category>
            <category><![CDATA[talent-development]]></category>
            <category><![CDATA[leadership]]></category>
            <dc:creator><![CDATA[Gurinderpal Singh]]></dc:creator>
            <pubDate>Wed, 05 Aug 2026 17:56:25 GMT</pubDate>
            <atom:updated>2026-08-05T19:00:47.207Z</atom:updated>
            <content:encoded><![CDATA[<p>A few years ago, I was sitting in a site risk register review with six people around the table. Four of us had been in sterile or advanced therapy manufacturing for more than fifteen years. The other two were early in their careers. That moment stayed with me, because I started seeing the same pattern in other places too: at conferences, in review boards, and across senior quality roles in biopharma. There are a lot of experienced quality leaders moving into the later part of their careers, and the bench behind them does not look as deep as it should.</p><p>I do not think we are talking about this enough.</p><h3>The pipeline problem hiding in plain sight</h3><p>Cell and gene therapy alone has created an explosion of manufacturing complexity over the last decade, new modalities, new facility types, new regulatory expectations, all requiring quality professionals who understand both classical sterile manufacturing discipline and the very different risk profile of advanced therapies. At the same time, the professionals who built their careers across that transition, who lived through the shift from pure biologics to viral vectors and cell therapy, are aging toward retirement.</p><p>The industry has spent enormous energy solving for manufacturing capacity. It has spent comparatively little energy solving for the talent pipeline that has to run that capacity safely. You cannot staff your way out of a shortage of judgment. Judgment is built over years, through mistakes, near-misses, and mentorship, not through a training module.</p><h3>What mentorship in quality actually requires</h3><p>Technical training is the easy part. There are courses, certifications, SOPs, and increasingly good formal education for the mechanics of quality systems, not to mention the progression through AI. What’s much harder to transmit is judgment: knowing when a deviation is truly minor and when it’s the first thread of something larger, knowing when to hold a line under schedule pressure and when the line you’re holding no longer matters, knowing how to walk into a room and be the person who slows things down without becoming the person nobody wants in the room.</p><p>That kind of judgment does not come from reading a manual. It comes from seeing how an experienced person thinks through an issue, hearing the reasoning behind the decision, and then being given the chance to make similar calls with support close by. That is mentorship. My concern is that senior quality people are already stretched across urgent operational work, supporting business and the time needed to teach this properly is often the first thing that gets squeezed.</p><h3>Quality has an image problem, and it is making the pipeline worse</h3><p>There is another reason the pipeline is thin, and I think we need to be honest about it. Quality has an image problem inside our own organizations. Walk into a room with engineering or manufacturing and say quality is joining the discussion. In many cases, the tone changes. People become more guarded. Some stop talking through the problem openly and start waiting to hear what quality is going to challenge.</p><p>I understand why that happens. In many situations, quality has been experienced as the function that shows up after the fact to identify what went wrong, rather than the partner who was involved early enough to help prevent the issue. That reputation did not appear by accident, and it will not change unless we change how we show up.</p><p>It also discourages some of the people we need most from considering quality as a career path. If a talented young engineer sees quality as the group everyone becomes careful around, why would they choose that path when operations or process engineering may look more collaborative and visible?</p><p>Quality also must be honest about the part we sometimes play in creating that image. It only takes a few people using the authority of the final signature the wrong way to make the whole function feel difficult to work with. When quality becomes rude, dismissive, or overly authoritative, people stop seeing us as partners in protecting the patient and product quality and start seeing us as the group with the power to block progress. That is a dangerous place for the function to be. The final signature should never be used as leverage. It should be the outcome of good collaboration, clear expectations, and trust built early enough that the decision at the end does not feel like a surprise.</p><h3>What senior leaders can actually do about it</h3><p>If we want strong people from manufacturing, engineering, and operations to see quality as a place where they can build a meaningful career, then senior quality leaders have to own part of that perception. We cannot expect the next generation to choose quality if we make the role look like a constant escalation point instead of a place where they can influence decisions early.</p><p>A few things I’ve found actually move the needle, rather than just sounding good in a leadership meeting.</p><p>Show up before there’s a problem to find. If the only time quality appears in a manufacturing area is during an investigation or an audit prep, we’ve trained people to associate quality with bad news by design. Senior quality leaders need to be visible during the boring, uneventful stretches too, on the floor, in design reviews, in routine huddles, so our presence isn’t itself a signal that something has gone wrong.</p><p>Ask before we tell. Walking into a room and immediately pointing out a gap trains people to brace rather than engage. Walking in and asking someone to walk you through their process, genuinely curious rather than auditing, does the opposite. It signals that quality wants to understand the work, not just judge it.</p><p>Make it safe to be wrong out loud. The fastest way to keep a room quiet is to punish the first honest mistake someone admits to in front of you. The leaders who build the most open relationships with other functions are the ones who visibly reward the engineer who says “I think we missed something” over the one who stays silent and hopes it’s not noticed.</p><p>Give credit publicly, correct privately. This sounds obvious and is rarely practiced. When manufacturing or engineering does something well from a quality perspective, say so, in front of their leadership, not just in a report nobody reads. Save the harder conversations for a smaller room. People remember which one happens in public.</p><p>Create real, informal access. Some of the best cross-functional trust I’ve built came from conversations that had nothing to do with an open deviation, grabbing coffee, sitting in on a planning meeting I wasn’t required to attend, being a familiar face rather than only an official one. If the only interface between quality and other departments is a formal escalation process, we shouldn’t be surprised when that’s how the relationship feels.</p><p>None of this lowers the standard. It changes when people bring problems to quality. Ideally, they bring them early, when they are still small, inexpensive, and easier to fix — not later, when they have become costly, visible, and much harder to manage. An organization where people hide small issues from quality out of habit is not more compliant; it is just more likely to be surprised later. In quality, almost any news can be managed. It’s the delayed news that becomes the real problem.</p><h3>The cost of getting this wrong</h3><p>If the current trajectory holds, the industry will have plenty of technically credentialed quality professionals and a shortage of people who’ve actually absorbed the harder, less teachable half of the job. That gap doesn’t show up immediately. It shows up five or ten years from now, in facilities being run by people who know the SOPs but haven’t yet developed the instinct for what the SOPs don’t cover, because nobody had the bandwidth to walk them through it.</p><p>I’ve seen what that instinct is worth. It’s the difference between catching a problem in a design review before it’s poured in concrete, and discovering it during an inspection years later. It is, in a very literal sense, the difference between a facility that operates safely for decades and one that generates a 483.</p><h3>What I am doing about it, and what I would ask of others in this field</h3><p>I have started being more deliberate about this in my own work. I am pulling early-career engineers into risk review meetings earlier, using those discussions to explain not just the decision, but the thinking behind it. I am also having more informal hallway conversations about their career goals and whether they have ever considered Quality as a path to the future.</p><p>I am working with my team to create more quality-focused events at the site and give people from other functions a better view of what quality actually does day to day. That includes opportunities to spend a few hours with quality, understand the real role of the function, and see how we support manufacturing, engineering, and ultimately the patient. We also need to be more willing to take a chance on strong talent from other departments. If we want the pipeline to improve, quality has to become more visible, approachable, and open to people who may not have followed the traditional path.</p><p>The facilities being designed and built today will be operated for decades. The people who need to be ready to lead quality inside them are, right now, sitting in rooms like the one I described at the start, quietly outnumbered. That is worth more attention than it’s currently getting, and it starts with those of us further along choosing to make the time, not waiting for the industry to formalize a solution first.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f258dfd84dd" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>