Tuesday, September 6, 2011

Over nineteen years experience in industry

What does it mean to have "over nineteen years experience in industry?" Here is the story of what I have done, told primarily with a focus on my accomplishments. This is a long and comprehensive post, and is intended to flesh out line items on my résumé, and explain how each item follows the last. It is not meant to serve as any sort of thesis, but instead provides a more detailed look at items that are summarized all-too-briefly elsewhere. Some analysis and lessons learned will follow in a later post, along with looking at these same experiences from other points of view.

Summers, 1986 - 1988: Rotating shifts in a TV faceplate factory

My first taste of the industrial world came in college, when I spent my Summers at the glass factory where my father worked. This was a factory that manufactured TV faceplates, and was one of the major employers in Northeastern Pennsylvania. It was hard and potentially dangerous work, hefting and inspecting heavy glass faceplates as they rolled through various stages of production. More difficult and dangerous than the work itself were the rotating shifts on which we worked: five days of afternoon shift, two days off, five days of morning shift, two days off, and then five days of night shift. (The actual schedule was somewhat more complicated than this, with occasional breaks of one-and-a-half or two-and-a-half days, and occasional sixteen-hour days of overtime.) It was hot, noisy, high-speed work that required strength and stamina and careful attention to safety.

(The company changed hands several times over the years when I worked there, and was eventually bought by a multinational corporation which closed the factory and took the jobs overseas. A few years later this all became moot as TV faceplates went the way of eight-track tapes and buggy whips.)

March 1990 - August 1991: Solar Cell manufacturer

After my brief stint in graduate school I found myself once more entering the world of industry, thanks to the assistance and guidance of one of the professors at my grad school. The company I found employment with manufactured solar cells, which seemed like an appropriate place for someone with a degree in Physics. The work itself was not something that called for a degree of any sort: hand-scrubbing silicon substrates that had gone through several processing steps on their way to becoming solar cells. After a few months I moved on to testing incoming substrate material and sorting it for later processing steps. It was around this time that I decided to renew my lease and stay in the area, to see if I could make a future with this expanding company. After another year that found me gradually crawling up the corporate ladder, eventually being given supervision of an entire sub-department dedicated to inspecting, sorting, and preparing incoming silicon substrates and assisting in the design of a brand-new facility, I concluded I was not living up to my employment or economic potential and decided to gracefully transition out of the job. I had learned a lot in my eighteen months there, and had had my first exposure to practices like "Bottom Up" management, the Kanban system, and Just-In-Time Manufacturing, but it was time to move on.

May 1992 - January 1993: Plater in CD Pre-Production department

After six months of nursing my temporarily bedridden grandmother back to health and enabling her to live independently once again, I managed to secure an interview with one of Northeastern Pennsylvania's largest employers, a record, tape, and compact disc (CD) manufacturer that had done some work with the Physics and Electronics Engineering Department back at the University of Scranton. The job that I interviewed for was in the CD Pre-Production department. The job itself was nothing much - it would be in the Plating department, using an electroforming technique to create the Stampers that would be used to manufacture CDs - but I was assured that there were things in the works that would call for the services of someone with my background.

After six months on a schedule of twelve-hour nights, 6:00 PM to 6:00 AM, four nights out of every eight with occasional overtime, the opportunity that had been hinted at became available. The department was looking for a Statistical Process Control Coordinator - someone with a strong background in mathematics who would be willing to learn and implement the techniques of Statistical Process Control. I jumped at the chance.

January 1993 - August 1996: Statistical Process Control Coordinator

When I took the job, it was still in the process of being defined; in fact, my partner and I did much of the defining. We were part of a team that reported to the Vice-President who was in charge of the CD Pre-Production department. That team - including the VP himself - traveled to Knoxville, Tennessee to take part in Dr. Donald Wheeler's Understanding Statistical Process Control seminar. This SPC boot camp was fundamental in directing the path we would take with the entire Pre-Production department.

This was also our first exposure to the genius of W. Edwards Deming. His management philosophies would guide us through many years of successful implementation of both the principles of Statistical Process Control as well as the set of practices known as Continuous Improvement. Our team would later take part in one of Dr. Deming's final telecourses before his death.

We introduced Statistical Process Control into every aspect of the Pre-Production department. This implementation followed a pattern familiar to anyone with experience with SPC: an initial dramatic improvement in yields, followed by a gradual stabilization as sources of assignable-cause variation were identified and eliminated and a further improvement as sources of common-cause variation - that is, the variation built into the system itself - were identified and minimized or improved.  Yields in the Plating department showed a dramatic improvement, from typical ranges of 80% - 85% before the implementation of SPC to 90% - 95% (and better) afterwards.

As we applied Statistical Process Control to more and more processes, the tedious but enlightening practice of creating and calculating control charts by hand became a bit overwhelming. We invested in a basic package of control charting software that was called SPC-PC IV, but I found it limited in its applications and a bit sloppy when it came to the actual rules of SPC. For example, it would routinely recalculate control limits based on all available data rather than compare recent data points to control limits established by an earlier set of data, to ask the question (all control charts ask questions - understanding that is fundamental to understanding control charts) "Does the recent data show the same sources of common-cause variation as the earlier data? That is, does the current process resemble the earlier process?" So I created my own control chart programs using the equations in Dr. Wheeler's book, first in Lotus and later in Excel. These were flexible and expandable, and could have control limits drawn based on a specified range of data and then applied to the entire data set. They served our purposes far better than the purchased SPC program, and gave verifiably correct results.

Statistical Process Control isn't something that can be implemented from the outside; it is something that must be integrated into processes, and understood by the people who operate those processes.  So after developing some expertise in the use of Statistical Process Control techniques, I created our own course in SPC that I could teach to the operators themselves. Initially offered to operators two or three at a time to minimize disruption of normal operations, this course was later modified into a two-day seminar (using Dr. Wheeler's Understanding Statistical Process Control as a text) that I presented to technicians, supervisors, and members of management in groups of twenty or more.

By this time the application of SPC was spilling out of the Pre-Production department and into CD Production itself.  Our group was now tasked with not only monitoring and analyzing performance data from within the processes that led up to the manufacture of CD stampers, but also with monitoring and analyzing the performance of the stampers as they manufactured compact discs. Our efforts in this regard were so successful that an assistant and I eventually traveled to our Los Angeles facility to install and customize our data collection and analysis programs (built on the dBase III+ platform)  there as well.

August 1996 - February 1999: Statistics and Database Management Technologist

It is said that there is nothing constant but change, and that is especially true in business. Management and leadership changes eventually steered our group away from expanded implementation of Statistical Process Control throughout the facility, instead focusing us on data collection and analysis. The formation of a new department focusing on the development of a new technology for digitally encoding video onto a disc format (which would eventually become known as "DVD") also resulted in the loss of critical personnel and the reconfiguration of departments.

None of this adversely affected the established SPC programs, and this time actually saw an expansion of SPC in those areas that were already using it. At the same time the sheer quantity of data coming back from processes was growing, and a management directive indicated that this data collection should become less labor-intensive. Initially data was gathered by having production operators jot down a few important pieces of data on a form, and then having the CD stamper library staff enter this data into a database, carefully punching in all identifying information along with the relevant data. Unfortunately, any small error with any part of this data entry could render entire database records useless without tedious review and correction of each record.

In response to this I developed and introduced several new innovations. Identification information for each stamper was already encoded using barcodes that allowed them to be scanned into and out of the library. Some quick research turned up an inexpensive bar code reader that could be used to convert this information into the input for a database field, minimizing the possibility of stamper misidentification. This sped up the process of entering data upon the return of the stamper. But then another idea hit me: why not eliminate this step altogether?

Stampers were sent out from the library to make discs. Discs were manufactured and tested at the presses, and the test results would be printed out. Press operators would then copy the relevant data from the test results onto the stamper ticket, which would be returned to the stamper room for data entry. What if we could have the tester itself store the test results directly in a format that could then be imported into a database?

It took several weeks of work with the people responsible for programming and maintaining the data racks, but in the end we had something sleek and very workable: identifying information for each stamper (and later, each press) would be scanned into the tester by the press operator, using the same inexpensive barcode scanners we had purchased for data entry at the stamper library, mounted to the testing racks using the sophisticated material known as Velcro. At the end of each test, the data from the tester would be written to a "flat file" and recorded to a 3.5" floppy disc. Periodically I would collect and replace the discs from all the testers and import the data into databases, and proceed from there.  The amount of data that could be recorded for each test exceeded the amount practically possible for any manual data recording and entry system.

The proposal was quickly approved by management, and after a few pilot systems were tested on the production floor, every tester in production was converted to the new system.  The new system resulted in a vast reduction in the amount of time needed for recording and entering data for each test, allowing press operators to increase their focus on production and quality and allowing stamper library operators to focus on customer service , while at the same time increasing the accuracy, quantity, and breadth of data available for analysis.

The data was not simply being collected for the sake of collection. We designed programs that allowed us to sort and analyze the data in numerous ways to look for trends and identify root causes of variation. By doing this we minimized performance variation actross the production floor, identified underlying problems, and used data-based decision making to improve yields.

It was during this period that I received training in the philosophy that was called Total Quality Management. It incorporated much of what I had learned about Statistical Process Control and Continuous Improvement, but added other ideas like Just-In-Time and Lean Manufacturing.

And then an opportunity arose in the newly-formed DVD department that I could not pass up.

February 1999 - February 2007: DVD Asset Manager

By this time DVD manufacturing existed not just as a separate department, but as a separate work unit entirely. After several years of development it had finally settled into a regular production system, or very nearly so. The department I was entering was analogous to the Pre-Production department I had started in nearly seven years earlier, and the work unit was led by the same Vice-President who had earlier spearheaded the SPC initiative.

My job would be to serve as the first completely dedicated DVD Asset Manager. The role of the Asset Manager would be to work with the customers/clients to identify all of the individual elements that would be present on the finished DVD project - every video piece, audio track, subtitle, photo gallery, ROM file, and everything else, take receipt of all of these different elements, or "assets," and then determine how they could fit together in the limited space of the DVD while maximizing the quality of each finished piece. I would not be the first person to do this job for our company; several others had done it on a part-time basis in addition to their other duties, but the amount of work was rapidly expanding to the point that a full-time, dedicated Asset Manager would be required.

After I accepted the job I began a crash course in the technical details of DVD compression, encoding, and authoring. At the same time I was also expected to smoothly transition out of my previous job. As I was now located in a standalone building with the rest of my new department, my workday involved a commute between facilities for the first few weeks.

I reviewed the tools that my predecessors had left for me. Projects were filed in boxes scattered throughout the department. The mathematical calculations that would go into determining the "Bit Budget" for each DVD layer would be done using a spreadsheet-based calculator and then painstakingly copied by hand into a custom-printed "Project Book."

My first innovation was a simple one: shelves.  My office would consist of a desk with a chair, a computer, and a phone, completely surrounded by shelves. And on these shelves would be the project boxes containing every DVD project we were working on, had recently worked on, or would be working on in the near future. Now we had a visual representation of our work in progress, our future workload, and quick access to recently completed projects. No longer would anyone have to ask, "Where's the project box for ____?" If somebody wasn't actively working on it, the answer would be: In the Asset Manager's office.

After that I began to hammer away at the Excel-based Bit Budget calculator. This was an excellent tool - but I didn't understand it completely. So I did what I always do in such a situation: I took it apart and looked at each part until I understood how everything worked. Then I put it back together again in a way that made sense to me. I expanded it, streamlined it, simplified it, made it more flexible and more customizable to reflect the numerous applications for different DVD projects.

And these projects ranged from simple to mind-manglingly complex, a multi-audio multi-subtitle feature split across both layers of a DVD-9 with motion menus and bonus content of various sizes and video quality, all of it also spread across both layers. The trick was to maximize the video quality of the feature (allowing it to have the highest possible average bitrate and an invisible layer break) while ensuring that the video quality of the bonus content remained acceptable and, whenever possible, was just as good as the feature. In truth, dealing with all of these variables at once called to mind my long-ago studies of the physics of complex systems.

I created other calculators for myself in Excel. The issue of timecode is a complicated one with DVDs, and we worked with three different timecode formats: drop-frame NTSC, non-drop-frame NTSC, and PAL. I created calculators that could quickly calculate the total number of seconds between any two specified timecodes - critical to any DVD bit budget - by entering a starting timecode, an ending timecode, and identifying the timecode format.  (This gets especially confusing when dealing with drop-frame timecode.)  These calculators also made it possible to quickly locate a specific timecode on a PAL video master from the corresponding timecode on an NTSC video master, and vice-versa. (This is a situation that often arises in DVD compression and encoding.)

As projects became more complicated I grew weary of the tedious hand-copying of data into the DVD Project Book like some monk in his chamber with an ancient manuscript. I revised the bit budget calculator to mimic the design of the relevant Project Book page, and streamlined my workload by simply printing and pasting the calculator page into the book.

As Asset Manager it was my responsibility to take receipt of all incoming subtitle files, and to prepare them for review by QC. Unfortunately, sometimes subtitle files would arrive mislabeled - Finnish might actually be Icelandic, or Danish actually German, or Japanese might be Korean, and it was up to me to spot the differences.  (Finnish uses a lot of double vowels, Icelandic has unique characters, Korean looks more boxy than Japanese, and Danish...well, I had a semester of German in college, so I had a fighting chance there.) Sometimes the English file would actually be English SDH (subtitles for the deaf and hearing-impaired, identical to the English but containing captions for sound effects and  songs, only used when explicitly called for by the customer.) Sometimes the files would reflect that they were built around a textless master, while the video master we were working from still had all "burned-in" text included. Sometimes the files had been built using a drop-frame timecode when the master we were working from was non-drop-frame, or vice-versa. In other words, there were many problems that could arise with subtitles, and it was my responsibility to deal with them, and to watch over and work with the people who were performing the actual subtitle QC.

DVD Asset Management wasn't all about dealing with assets. I also had to serve as the secondary (or, in some cases, primary) contact with clients. And these clients came in all sorts. Some worked for major studios, and their job was essentially the complement of mine; they knew precisely what a job called for, and precisely when each asset would be delivered. Some worked for smaller studios, and were virtually one-person-shows (although most of them had assistants); in these cases there was a lot of making-it-up-as-you-go-along with each project, sometimes with final content being determined on the day the project was due to be sent out from our department. Some clients were the actual directors or producers of the features being turned into DVDs, and sometimes the stars as well, and in these cases there was a learning curve on both sides of the relationship. But in every case it was my responsibility to provide the highest level of quality and customer service in the shortest turnaround time at the lowest total cost. And our department truly performed in this regard:  When I arrived the standard stated turnaround time was "six weeks from receipt of last asset"; within a few years, through tremendous team effort, we had brought this down to less than six hours from receipt of last asset.

Project Box storage and retrieval was another of my responsibilities. This was done in a fairly chaotic manner in the building that housed our department when I joined. When we were informed that we would be moving into a new location in the main production facility, we were given the opportunity to design the department to fit our needs. The new storage location was more spacious, but the filing of assets took on a chaotic form once again as projects from numerous clients continued to roll in. Finally I instituted a consolidation and asset return process that resulted in increased availability of storage space for future projects, and allowed for a more methodical organization of remaining project boxes.

Unrelated to any of my other tasks as Asset Manager, I was also assigned the responsibility of maintaining departmental ISO compliance. This was my first real-world exposure to ISO, so it was a definite learning experience.

One of the chronic annoyances throughout my time as Asset Manager was the Project Book. These books had been designed before I arrived, and were revised at least once after my arrival, but problems persisted. They were bulky physical books designed in a one-size-fits-all approach. Most had vast expanses of content that went blank in almost every project, while in other cases additional pages had to be photocopied and stapled in. They could be filled out illegibly or not at all, lost, filed in the wrong project box, accidentally destroyed, or accidentally thrown away. Every project needed its own book, even projects that were identical except for a single menu or a few subtitle languages. And each one cost quite a bit of money, and had to be bought in bulk.

Eventually the decision came down from above that we had to cut expenses, and the project book was low-hanging fruit. We decided to replace it with an electronic version. But almost immediately we ran into problems. Working with our IT department meant that we would have to get onto their schedule of projects and be assigned a priority - and our priorities were not necessarily their priorities. From an earlier attempt at doing something similar, I knew that the approach from our IT department would be to create a nuclear-powered nutcracker - create a solution that could easily deal with the most complex projects, but then require that overhead to be carried around on the simplest projects as well.

After several weeks of fruitless attempts to get the ball rolling, I announced that I had figured out a way of doing this that would not require IT assistance. I could have it together in three weeks, though I would require input and cooperation from everyone in the department.

I arrived at my solution using the Feynman method: I thought about the problem very hard and then envisioned what the successful solution would look like. The rest was just tedious details that would need to be worked out over the next two or three weeks. I envisioned an "e-book," an electronic book consisting of a single multi-tabbed Excel spreadsheet for each project. I started with my existing Excel spreadsheets for the bit-budget calculators - which had already been designed to mimic the look of the Project Book - and worked my way out from there. I added in some timecode calculators, mostly for my own convenience, but then reconfigured each bitrate calculator to incorporate a timecode calculator - and was even able to toggle between drop-frame, non-drop-frame, and PAL timecode with some simple IF statements. I reconstructed the front page of the Project Book, which listed all of the assets for the project, and then linked it to the bit-budget calculator to automatically incorporate audio and subtitles specified on this page into the bit-budget page and the calculations. I worked with every sub-department within our department to determine their needs as well as their wants. The goal was not simply to reproduce the Project Book, but to create the ideal project book that would address all of our needs and could be easily customized to each project.

And in the end it worked. By the end of the three weeks the Excel-based e-book had been designed, tested, revised, and re-tested, and was ready for production. It was a complex, multi-tabbed spreadsheet where data entered on any given page would be reflected in relevant locations on other pages, even automatically activating calculators where applicable. It would be stored on a shared drive that could only be accessed by computers within our department. It could be opened exclusively for editing by one user at a time, or could be opened in read-only mode by multiple users simultaneously. It eliminated much of the tedium of filling out the Project Book, and made creating bit-budgets a breeze. It couldn't be lost - at least, not without considerable effort - and it could be used by people scattered across the department simultaneously. Illegible handwriting was no longer an issue. And each new book cost nothing to create. In terms of quality, customer service, turnaround time, and cost, the new electronic Project Book that I had designed had it all. It is, as far as I know, still being used to this day.

Perhaps it made it a little too easy to now do bit-budgets, so easy that no specialized skill or knowledge was called for. After having endured several lean years due to a downturn in the industry and increased competition from other DVD Compression, Encoding, and Authoring facilities, our department sustained several rounds of staffing cuts, taking us to the bone and worse. And finally, at the end of February 2007, as part of the most drastic departmental staffing cut yet, the position of Asset Manager was eliminated.

August 2007 - December 2007: DVD Press Operator

After six months of layoff, I received a call from the company's HR department and was offered the opportunity to return to work, retaining my seniority and benefits as though there had been no interruption in my service. The only catch was that I would no longer be a salaried employee, as I had been since I took the position as SPC Coordinator; I would have my choice of becoming a DVD Press Operator, or a DVD Print Operator.

By now the company had become primarily a DVD replicator, just ten years after the format had been invented. Any other task - Pre-Production, Printing, even the tasks my previous department had done - could be outsourced and the company would remain. But basic DVD replication could not be outsourced without outsourcing the primary function of the company itself. I chose the DVD Press Operator job.

DVD replication was a real eye-opener. When I left the world of production behind in 1999, DVD technology was still something new. By 2007 it was old hat, routine, just another set of production units in the factory. But it was all new to me. I handled stampers for the first time in nearly a decade - DVD stampers instead of CD stampers, with much smaller pits and much tighter tracks - and learned of the extremely precise techniques that went into installing a DVD stamper in a press, where the smallest fleck of dust could easily destroy a stamper or, if not noticed immediately, ruin an entire production run.

I also learned how to run DVD presses. I have never considered myself particularly mechanically inclined, but now I had to learn how to operate several different designs of large, potentially dangerous injection molding machines, which worked at high temperatures and devastating pressures to pop out dozens of molded DVDs each minute.

It took a while. There were so many details to attend to, and all of them were important. Nothing could be skipped over without risking quality or safety. I never learned to operate as quickly as the experienced operators who made everything look easy, but I was eventually able to fly solo and operate my own groups of presses.

DVD press operation wasn't just about installing stampers and molding DVDs; we also had to learn how to run electrical and physical quality tests and to interpret test results and adjust the processes in ways that would bring measured parameters within acceptable ranges. As each process adjustment could affect several different parameters, and multiple adjustments could have multiple interactions with each other, this became both an art and a science.

I powered through this job for several months. The pay was considerably less than what I had been making as DVD Asset Manager, but it was considerably more than what was being offered by those few jobs that were available elsewhere. But after less than six months, I was tapped for a special project - something that would call on skills I hadn't had to fully use since before I became the Asset Manager.

January 2008 - December 2008: Data Analyst

A renewed interest in data-based decision making had arisen in recent years. Two salaried employees - both managers - had been pulled away from their regular jobs to collect and analyze data and assemble it into reports that would be presented at morning meetings each day. With my abilities known, the decision was made to have these tasks re-assigned from the more highly-paid salaried managers to me.

I was in a bit of a pickle. I was taking over reports compiled by two individual people. In the week that I was given to learn their techniques for gathering the data and creating the reports, I realized that each report required one individual working at full-speed for the entire time leading up to the delivery of the report. And I had to handle both of them. So I had to find ways to get both reports out faster, without losing any of the data.

I did the same thing I had done many times before: analysis and synthesis. Taking apart the problem, studying it, at reassembling it in a way that made sense to me. I quickly redesigned each report in a way that was identical in its outward appearance, but allowed for more efficient data entry. Then I turned to the sources of the data, which had to be tediously assembled from several other reports and the outputs of several programs. I identified ways of increasing the efficiency of this process, streamlining the source reports and manipulating their outputs into formats that made for easy data input for the final daily reports. I also improved the accuracy of the data by contacting the parties responsible for the compiling of these reports and working with them to eliminate several sources of redundancy, ambiguity, and outright error.

In the end I was able to strip down, redesign, and re-assemble these reports so they could be compiled far more quickly, with a higher confidence in their accuracy. What had required upwards of six man-hours each day could now be done in two. And I could turn my attention to other issues.

I once again had access to and responsibility for a massive flow of data. Where once it was CD stamper performance at press data, now it was DVD press yields. I was tasked with applying my special abilities with data to these data sets, making them dance and sing the way I had once done with CDs.

On top of this a new technology was arising, so new that the format - and the name - had not yet been agreed upon.  Two competing formats were being produced: HD-DVD, a high-definition version of the DVD format, and BD or Blu-Ray, an entirely new technology that had very little in common with DVD beyond the shape of the disc. One of my tasks was to analyze and report on the performance of the various systems that had been bought or modified to make the new BD and HD-DVD formats.

Unfortunately, as the year went on, things changed again. I was still a DVD Press Operator, and I needed to be returned to the pool of DVD Press Operators. As November drew to a close I finished up my work of data analysis, once again smoothly transitioning my reports to another individual.

December 2008 - December 2010: DVD Press Operator

Upon returning to the DVD press floor I was almost immediately laid off. Rolling layoffs had been a reality for some time, and my Data Analyst position kept me immune for a while. But as soon as I resumed my press operator duties I was also subject to periodic layoffs.

One of these layoffs stretched on longer than others, and I found out our company had gone through one of its occasional reductions in force. I was not one of the victims of this permanent layoff, but I found that due to losses from other shifts, I would need to be reassigned, for the first time in over sixteen years, to night shift.

I took to it. I adapted to night shift quickly. Adjustments to diet and sleep schedules came surprisingly easily. I threw myself at night shift, worked as much overtime as possible, and established working relationships with my new co-workers on night shift.

But more changes were in the works. Industry-wide changes, and changes to our company. We were given advance warning, but also had reason to hope that the worst would not come to pass. But it did anyway, for some of us. In December 2010, many of my co-workers and I experienced one final layoff.

Which brings us to here.

No comments:

Post a Comment