In-house translation – outside the box

Tag: in-house

  • 7 thoughts: learnings from an offline data room of raw MT

    7 thoughts: learnings from an offline data room of raw MT

    It has been a busy few weeks. My project and process management experience have been put to the test in a different manner. I had to process large volumes of machine translations for subsequent access in an offline data room as part of an inspection mission. Extensive use of raw machine translation was the only possible approach: the quantities of translations required for gisting purposes vastly exceeded available internal and external human translation capacity. In some cases required documents could not be machine translated for various reasons. I would have suggested that “pigs might fly” if anyone had suggested anything to the contrary.

    As I have mentioned elsewhere, in some instances, there was also the risk that raw MT output could result in creating unnecessary downstream workload. Fortunately, such unnecessary work was widely avoided by means of effective but sparing use of human translation. Human translation often provided clarity where raw MT output could not. The exercise also proved useful in providing insights about non-translators’ perception of translation. It reinforced my understanding of the reality for consumers of translation.

    This post addresses several points left unanswered in my first “7 thoughts” piece in January 2025.

    1. Translation remains a blackbox for many non-linguists: their sole interest is a passable output their input. They are not interested in the procedural steps, particularly the obscure ones that assist in ensuring quality. However, issues do surface about the rendering of key terms. This opens up discussions on the need for terminology. The issue has surfaced in my M365 Copilot Champions training. In our icebreaker “speed-dating” sessions with other champions many said they use Copilot to translate. I’d usually follow-up and take them up on issues encountered: the issue of inconsistent terminology leached out.
    2. Having machine translation on tap, seemingly at no charge, can make people wasteful of resources: “seemingly limitless” amounts of capacity available for machine translation meant inflated amounts of translation, than had only human translation capacity been available. Similarly, as meetings expand to fill time slots, no limits on submissions mean that people are less disciplined in the amount of files they request for translation. We lose sight of what is truly essential. In some cases submissions were entire electronic files (with 100s of files and 1000s of pages). I added notes about existing human translations that I supplied as an oversight mechanism. This sometimes elicited surprise: people were unaware of translations’ existence. I’ve taken this away as feedback for the future regarding marketing my output internally.
    3. “Good enough” is really all some people are after: in some cases, a raw machine translation exceeds the requesting party’s own capacity for a “translation”. Logically, such users will embrace machine translation in this case as the default option. Seductive fluency of Modern NMT and LLM/GenAI-based systems also reinforces their satisfaction with the output. Similarly, there is greater tolerance in terms of expectations for MT output over human translation. This issue demonstrates what the translation profession is up against: it is difficult for some to charge premium rates for a time-consuming process, where near instanteous raw MT output apparently “satisfies” their need.
    4. MTPE has a different cognitive load to human translation: I took care when setting out my stall. I chose to centralised rather than delegate handling the documents. By centralising the task, I had a full set of documents in case I needed them. I chose not to look at the raw MT output: it was available on a strictly “as is” basis. This was the only possible option: checking them would have resulted in the need to knock them into shape. That was not my task. Indeed, I would probably have struggled to have found a happy medium in post-editing. Inconsistent terminology would quickly have become an annoyance. And I needed a strategy to do this extra task while not affecting my own working processes, to cement that I work as the expert in the lead and remain firmly in the driving seat. I would not want to suddenly be the passenger, a “human in the loop” in the translation meaning. I became ruthless: discovering limits on daily limits on numbers of documents, size of files, numbers of files, Translating 1,000s of files, while not losing my own translation time was essential. I’d fed batches of files to translate while I was eating lunch, in a call, or waiting for seminars to begin.
    5. The advantage of centralising the process with a translator: given that machine translation is such a black box, a translator needs to maintain an oversight or even override role. In the worst case, I knew how the machine translation had been conducted (i.e. engine/model used, source format etc.) In the case of a poor translation, I also wanted the option to run the source text through the MT engine and have a bilingual XLIFF or TMX output as a starting point for a re-translation, without needing to align the text myself. This was also essential for ensuring the strict segregation of human and machine translation. Also this meant that I had to receive source texts, to avoid being inundated with with target texts with no source text, e.g. texts of GenAI produced output to correct.
    6. The grey zone about internal reliance on machine translation is greater than previously realised: NMT and GenAI have increased staff members’ general awareness of solutions like DeepL (put on a pedestal over Google Translate). The same can be said of the putative ability of GenAI/LLMs to translate. This awareness leads to far more “grey” machine translation than thought. Staff members will use any available solution unless it is physically blocked. Some institutions will actively block tools like Grammarly due to security risks entailed by using it. The same is true for LLMs/GenAI. Until a company blocks access to ChatGPT or DeepSeek etc. staff members will still use them. Merely having rules prohibiting such solutions is not a sufficient deterent.
    7. Effective outcome-based human translation and poor raw MT output can reinforce people’s confidence in human translation: despite seductive fluency, and “passable” quality and speed of delivery and availability, the value of human translation can shine through when the output really matters. People still value knowing they can rely on a translation. Some jobs remain too sensitive for MT despite mitigation measures taken. The human touch also shines through subtly. In one case I had to translate two iterations of a presentation (i.e. to show the progress made on a project).
      I translated the files together (using a virtual merge). This helped ensure the consistency of terminology, which helped guide discussions. MT struggles to be as consistent (e.g. repeated segments inconsistently translated across documents). Then I added a personal touch – I added a “virtual sticky note” to the cover slides to provide a reference to the abbreviations used. By eliminating the guesswork for the reader, it also impressed that this was a human translation from the outset.

    How did I go about it:

    I took a calculated approach to avoid unnecessary cognitive downstreaming to me.

    • I did quick checks for all received folders for standardised documents with only differing content (e.g. a reused template). In certain cases I made template legend available as a human translation in addition to raw MT.
    • On occasions I translated lists e.g. of process names. In preparing the translation, I chose an approach to prioritise maximising meaning over concision – and to ensure that related documents also used these process names – to help improve coherence across many documents.
    • I prioritised document requests arising from meetings for human translation, and highlighted that they were provided as human translations rather than MT.
    • I explained internal jargon (e.g. names of certain committees, working groups, names of soft law instruments) – e.g. using the virtual sticky note.
    • I prioritised some brief softlaw instruments for translation and publication, thereby increasing the permanence of the translation, in contrast to Raw MT for “here and now”.
    • I set up workflows to allow quick alignment of MT and separate read only TMs for MT output, to avoid MT translation contaminating working TMs, while also providing a starting point for human translation.
    • I used feedback requests from specialist divisions to add terminology, and locate documents prioritised for future translation, revision and checking by subject matter experts and publication.

    Turning the process into a pipeline for ongoing translation:

    The successful conclusion to the mission should not be the end of the story. There is another mission ahead. In this regard, the next steps are:

    • Giving contributors and coordinators the chance to find out what went well, and what we could try to improve.
    • Adding a chapter to my language services handbook (LaSH) based on the learnings from the feedback.
    • Prioritising a list of translations for publication prior to the next mission.
    • Process optimisation
      • ensuring documents for translation are saved separately and not embedded in other files.
      • Nomenclature rules for filenames (including limiting length of filenames.)
      • Handling of large files (e.g. chunking to reduce waiting time or uploading for translation during breaks etc.)
  • 7 thoughts on the challenges facing in-house translators

    7 thoughts on the challenges facing in-house translators

    At the start of this week, I attended an event organised for in-house translators by Universitas. I was part of a small and eclectic group, meeting at the end of the working day. From a very interesting and relaxed couple of hours talking with the others in the group a number of things sprung to mine, which I addressed the following day on LinkedIn.

    1. There is no “one-size-fits-all” in-house role. Expectations on in-house translators vary dramatically: from subject matter specialists through to multi-language pair generalists. There is no hard and fast rule about work in a single language pair or multiple pairs. Even classical “Translator” positions may involve a mixture of translation and other activities. For applicants for such positions, it is advisable at interview to ask about the likely expected hours dedicated to translation.
    2. Job descriptions for open vacancies are seldom as straightforward as “Translator (m/f/d)”. Translation is frequently only a part of the job description, and the job title is seldom for only a translator. In language career portals in the German-speaking world (e.g. Stepstone, Kununu or Karriere.at) most hits for the skill “Übersetzung” talk about it figuratively.
      For example a recent job advert for Österreichische Post AG for an Expert in Controlling Insights mentioned “Du fungierst als “Dolmetscher” zwischen Fachbereich und Programmierer und bist zuständig für die Übersetzung der Anforderungen des Fachbereichs in detaillierte Vorgaben und Zielsetzungen für technische Umsetzung.” In other words – nothing to do with translation or interpreting! In other cases it might be disguised in a job description for a “zweisprachige Schreibkraft” or “Kommunikationstalent”. This latter role is closer to transcreation than translation. Here are some common language-based job titles.
    3. The burgeoning TechStack: among the group around the table, the tools used, and expectations regarding such tool use was varied. There were also varying views about the expectations regarding the use of GenAI/LLMs. Some are very open to the possibilities, while others actively decline to use such tools, or are not permitted to do so.
      Often required training regarding GenAI/LLMs is not specifically tailored to translators. Similarly, the effective use of such tools for translators is not as clear cut as the hype makes out. Deployment of new tools is often IT-led. This approach sometimes overlooks those with genuine expertise to really use them and assess the quality of their output.

    Resource issues

    1. Double-hatting is common: some translators also work as interpreters, rather than having separate translation and interpreting personnel. Others in-housers are only part-time in-house, so have to juggle self-employment alongside their fixed employment.
      This naturally places additional demands on them in terms of time management and also how to organise their time effectively. Teleworking may have cut out some unnecessary miles/kilometers, but there is still a lot of juggling required with multiple positions.
    2. Working for a demanding customer base with dwindling human translation capacity: this can become even more difficult if language services are overseen by non-linguists. This can make it more difficult to discuss the need for quality that goes beyond “good enough”, and where “fit for purpose” is a minimum requirement.
      It can be difficult to get past only being viewed as a cost centre. Translation can be quantified easily in terms of cost, but its impact on sales etc. is more difficult to quantify. Tighter budgets mean fewer retiring colleagues are replaced, or FTEs are replaced by fractional headcount. Alternatively FTEs might only have a certain percentage of their time devoted to translation.
    3. Decreased job security: even in public administration there are perceptions that the job security of translators is lower than it used to be. The erosion of the classical triple constraint, the rise of “good enough”, and the improved “linguistic fitness” of many white collar colleagues has affected demand for translators.
      Job mobility and exchange programmes while studying mean that many colleagues are more confident in their language abilities that only a few years ago. However, there is still a subjective basis to their assessment of their own language abilities. Just as having two hands and a piano does not make me a concert pianist, working knowledge of two languages, does not automatically transfer into being able to write well in your target language.
    4. Rising expectations in terms of output: while tools like CAT and (N)MT have helped to increase translator productivity, there is still the unrealistic expectation in light of the promises of “instant translation” offered by browser-based tools.
      Translators’ potential output can really depend on so many factors – NMT/GenAI/LLMs are “confidently wrong” – they will always offer a translation, whereas the “cautiously correct” human translator reverts to the author if unsure – to clear up potential source or target text ambiguities.
      Similarly, expectations vary wildly based on the percentage of time spent on translation compared to non-translation activities. Often there is no dedicated capacity for terminology work. Only larger language units have dedicated terminologists: without them, it is often widely neglected. With the advent of MT/GenAI and the Terminology Augmented Generation approach, which is used to import your terminology into the LLM, it is likely to gain in importance.

    Are you interested in events like this? Universitas holds regular events throughout the year. Check out the Universitas website for more information – if you are not yet a member, some events are open to guests. If you are interested in knowing more about what I do, then why not join the Universitas Berufsbilder webinar on 23 October 2025, which will focus on the role of project management and process management.

  • Busting the 100% productivity myth: great(ly exaggerated corporate) expectations

    Busting the 100% productivity myth: great(ly exaggerated corporate) expectations

    A post on LinkedIn recently addressed the issue of expectations for delivery of a translation project. The suggested timeframe provided for a single translator to translate a website of approximately 25,000 words was approximately 1 week. The responses of other linguists generally fell into two distinct camps: firstly, the that’s-no-way-near-enough-time camp, and secondly, the it’s-no-wonder-translators-are-losing-out-to-MT-if-they-are-that-slow camp. Fence-sitters would probably fall into a how-long’s-a-piece-of-string camp – which is a justified argument – as the subject matter was unclear.

    Currently there are more “famine” than “feast” posts from freelancers. (N)MT and LLM-based translation form a two-pronged attack that are affecting human translators. Industry-side evangelisers sometimes claim that MT more content translation than human translators can translate. Even if this is the case, there is still a diminishing wedge for human translators.

    Since 2022, I have regularly seen posts about translators being reduced to post-editors of Machine Translation. The rates do not reflect the true amount of effort required to bring translations up to standard. Which in turn leads to a drop in motivation. It isn’t realistic to expect the same service for a living rate as a dumping rate.

    100% productivity is corporate settings: an illusion

    In the modern data-driven world, we are incredibly IT-dependent. Updates need to be done, and they don’t always happen overnight, during lunchbreaks etc. I’ve previously covered why I schedule my return to work to allow me to start with a home office day: with a “soft logon” the night before. Unless you user blocker appointments, you are bombarded with mails, calls, Teams chats etc. And all this eats into your productivity – particularly if you consider your day like a game of Tetris.

    As I pointed out to one comment about the 25,000 words in a week, which suggested 100% productivity in the corporate world, this is a fallacy. Time and activity tracking frequently sanitises out “Tür-und-Angel-Gespräche” with colleagues, lunchbreaks that overrun, online calls that start and end late. Full calendars are seldom a sign of productivity in their own right. There are also “meetings that could have been a mail” and continuous calls are draining. I now maintain better call discipline – sticking rigidly to the intended call length, and excusing myself from over-running calls.

    Is human productivity the issue?

    Returning to the how-long’s-a-piece-of-string issue, about productivity and its effect on translation output, it is clear that there are unreasonably high expectations on productivity. As a translator, you might have a “straightline top speed”, but for how long can you maintain it for? And does the ride remain comfortable, or do things straight to rattle or get uncomfortable. When I went in house, to try to gauge my output, I set myself an original 1,500 words a notional daily output. A 1,500 word document to translate from scratch can reasonably be expected to be sent back by the end of the day,

    Would I start translating the second I got into the office? Rarely. Unless an item has come in the previous evening and I had set up the project the previous evening. It might be necessary to perform some alignments, concordance-based terminology work, or (re)read the legislation. Sharing an office means inevitable phone calls and distractions. I often work with noise cancelling headphones when the office is fully occupied. When I have a lot of short tasks I use desktop timers to keep moving between the tasks.

    6 out of 8, or 8 out of 10?

    If I am lucky, I get about 6 hours (out of 8 hours) undisturbed translation time a day, and would have to go at a steady 250 words an hour to do 1,500 words in that 6 hours. As translation memories and termbases grew, “plain vanilla” translations became a lot quicker. Filler tasks like translating investment warnings are now practically automated. The translation task mainly involves locking a few segments and a quick check of the output and a bit of formatting.

    Consequently, I have been able to increase my notional daily output to 2,000 words, but the added 500 words a day reflect a number of factors:

    • I do considerably less terminology work. Now it is frequently ad hoc rather than in dedicated terminology sessions.
    • I also have read-only translation memories containing bilingual alignments of European law at my finger tips, allowing me to spend more time in Trados Studio than I previously did.
    • Better screen setup means reference materials open on a second screen, a glance away.
    • I have a very narrow subject focus – at its broadest, my subject matter is financial market supervision, but predominantly focussed on banking supervision. There are very few supervisory procedures that are genuinely new. I have occasional forays into insurance and Pensionskassen supervision, securities supervision or banking resolution.
    • Regular expressions for QA have helped reduce cognitive (over)load.

    Despite such “efficiency” improvements, achieving 8 hours’ pure translation productivity still requires working for over eight hours. Changes in daylight conditions also need considering. However, mature TMs also have drawbacks – which is why I have looked into better use of segment penalties, and terminology can also change over the years.

    Barnes’ Iron Triangle applied to translation

    As I alluded to in a previous post about imposter syndrome affecting translators – and how I banished my early career doubts, unrealistic expectations from customers are a genuine problem. For translation, the holy trinity of specifications consist of price, quality and time.

    triangle showing quality, price and time, to illustrate Barnes' Iron Triangle.

    Explained simply, it goes like this. You want a high quality translation? You’ll either have to pay a premium rate (i.e. price is high), or allow more time for the translation. You want a quick translation? You’ll either have to pay a premium rate (i.e. price is high due to needing translators to work extra hours, or in a team) or sacrifice quality. You want a cheap translation? You either sacrifice quality (e.g. review processes, terminology checks, coherence checks) or have to wait on delivery.

    The AI hype and the genuine advances in machine translation have pitted the industry against the professionals. There is a different playing field in the age of NMT and generative AI. There has certainly been a big leap since statistic MT was in its heyday. You have to therefore manage your customer’s expectations (explain what you do – e.g. explain that you use CAT and not (N)MT), and what the expected delivery time is.

    Managing expectations.

    I’ve always believed in expectation management (a skill you learn as a parent). Back in 2016, along with recurring daily work, we had most substantial relaunch of my employer’s website to date. Eight years on, there are still regularly new pages and posts, and the workflow has proven itself. I had to work to a fixed deadline for go live, at the end of an intense month (including work trips to London, Zagreb and Nuremberg).

    The project allowed me to also educate colleagues/customers about realistic expectations, while also changing the translation workflow for publishing directly to the website. Now, with backend CMS access. I extract texts from the source view in the CMS and open the files in Trados Studio. I could translate pages as they successively went live in the testing environment. That approach eliminated dealing with multiple versions of the same page or post as Word files. Agreeing on a top-down approach allows prioritisation of certain content for translation. This ensured handling top level content child pages/posts first, and steadily working through subpages.

    For multi-day projects, I explain how to involve me before a final version of the document exists. This approach is particularly useful for multiple iterations of a text. It also helps to allow more translation time – PerfectMatch helps to overcome document iteration issues. Naturally, I do also make sure that I allow a slight buffer, and early delivery is easier than having tight deadlines.

    Ultimately good customer communication is key – keep they updated about progress – maybe check in with them partway through the project – possibly the earlier the better. Try to group questions about terminology or wording suggestions together rather than a constant trickle of questions.

  • How in-house translators spend their holiday(s)

    How in-house translators spend their holiday(s)

    I turned off my work laptop after submitting my timesheet and activity record, and set my out of office message. Holiday had felt overdue for at least the last two weeks. August commutes had been punctuated by the slew of articles about “quiet quitting” – this summer’s controversial buzzword. That was until I ditched my smartphone on the tram and bus for my Kindle. “Taking back control” they call it. In another way to those who decided to take back control on holiday by taking stock of their situation and initiating change.

    Flagging mask-wearing attitude in Vienna indicated that people consider the Covid-19 pandemic is over. When commuting on public transport, between one-third and one-half of passengers ignored the FFP2 requirement. Although triple-vaccinated and recovered, the final two weeks were in fear of a new infection before my holiday.

    I had no thoughts of “quiet quitting”, but what was on my mind in ahead of my holiday to Tyrol, Salzburg and Bavaria?

    Before physically setting off on holiday, I powered down and guided my eldest son through his final days at Kindergarten. He clearly also needed a break. Before we left, I packed and loaded the family car, and finished some admin to allow me to relax. A catch-up with a friend from Uni and his family over beers was a perfect wind-down exercise.

    As I was leaving for my holidays, freelancer contacts announced their return to the office after theirs. They posted about marketing and customer acquisition work, due to a dread about a lack of work. It was similar when I freelanced, even with a steady set of fixed customers.

    So what was/is on my mind?

    Years in the same position have reassured me that I can go on holiday, relax as intended and take a third week off that is crucial for regeneration. Focussing on relaxation and regeneration, I downloaded several books for my Kindle, to enjoy while away. In addition, I stocked up on enjoyable podcast episodes (Decades from Home was a shoe-in given our visits to Bavaria). I also dialled down my current affairs intake – so took a break from Today in Focus. And deliberately also didn’t start getting into The News Agents until my return from holidays. My reading list consists of several biographies, some non-fiction reading and a couple of books related to translation.

    Sufficient holiday reading has become essential, especially when booking issues mean sharing a room (and large bed) with your sons. My wife shares with our daughter, when requested adjoining rooms are overlooked. The boys flake out early, giving me a couple of hours’ undisturbed reading. Due to their nocturnal thrashing about, I often end up sleep at the bottom of the bed.

    Holiday days vs being on holiday

    Taking paid leave (holiday days) is vastly different to “going on holiday”. Out of three weeks paid leave for my summer break, only half of that is actually spent on holiday. I assign a day of paid leave at the start to catch up on filed admin tasks. This quick “stock take” clears my mind for relaxing while on holiday. And I make a “before” list. Similarly, I have a “soft return” to the office in that I log on from home the evening before starting back to get the software updates done and come up with my “after” list. It is my way to ensure that I resume “on the ‘B’ of bang“.

    Comparing the “before” and “after” lists helps me see whether my pre- and post-holiday thoughts are on the same wavelength. They aren’t always – some “tired” thoughts at the start of my leave period are duly ditched. A third “during list” contain the thoughts that flitted through my mind while away (relaxation inspires!).

    I first became aware of the difference between holidays and being on holiday in 1994. It was after my first year of A Levels. My teachers gave me holiday reading in French and German. One explained with a smile that school holidays didn’t mean a holiday from learning. Surely paid leave is a different matter though?

    Does switching off the computer mean really switching off?

    While I switch off from my work as an in-house translator, I still struggle to switch off completely. I do try and reduce some of the sources of interference. This year, it was actively leaving all Xing groups. The step was to pre-empt their closure. Doing so also gave me instant impulses for my “during list”, which in turn help with my “after” list.

    As a translator, I actively avoid looking for mistranslations in restaurant menus on holiday. I know that non-translator friends will supply enough source material to make up for me not looking. And some still think I need to see the culinary howlers. I now untag and delete these kind of posts. I do this to actively stop thoughts wandering to ongoing and future translation projects.

    Breaking the (social) media onslaught

    One reason I choose active holidays (in terms of going out and doing and seeing things, rather than lying on a beach or a sofa or a bed) is that it helps me reduce my consumption of other social media. It is also good for actively being present for my children, seeing and experiencing the world. I mute Facebook and WhatsApp group chats (in particular WhatsApp parents groups as our three were away from Kindergarten) to reduce distraction.

    I also partially mute Twitter (and actively add some words to the block list!) particularly to avoid its toxicity. Instead, I actively reach for the Kindle and read when tempted to doomscroll. As mentioned above, I also limit my news consumption – in desperate times of rolling news, it is necessary to take a break from it all. It also helps avoid the emergence of a holiday fug by removing the everyday stimulus.

    Does anything meaningful come out of my lists?

    I mentioned having three lists (before, during and after) during my leave period. My “before list” had three to dos that I completed once back from holiday. Satisfyingly, I didn’t think about them once while on holiday. Two work-related items were also handled in my absence.

    My “during list” was an interesting one. Thoughts while relaxed on holiday, hastily scribbled on a free postcard from a bar, might develop into tangible projects. I came up with a machine learning use case for a forthcoming course while sat drinking a can of Spezi. Two conferences I want to attend (preferably in person!) flitted through my mind. I had a crazy thought involving regular expressions (RegEx) that could bear fruits. And there was a mental plan how to rearrange the office at home.

    My “after list” prepares me for the return to office. I log in once in advance to let my computer update before I go back to the office. I also catch up on calls for papers for conferences. This is a positive sign, as I recently decided not to submit abstracts, nor to attend a couple of conferences. It also contains a very quick list of all the mails from a single session in Outlook that I need to attend to. I usually revise that part of the list to make sure I prioritise what needs dealing with first.

    How do hybrid working arrangements optimise returning to work?

    As mentioned elsewhere, I work from home on Wednesdays and Thursdays. I therefore try to ensure that my holiday weeks (of five consecutive working days) run from Thursday to Wednesday or Wednesday to Tuesday. This means that I start back from paid leave in my office at home, and can get on with work with less disturbances than if in the office from day one.

    Similarly, when preparing to go on holiday, I try to ensure that I finish in the office one day before I have my final working day. It means that I think more carefully about what I need for the final working day, and focus on what needs finishing off. It also means that I can leave my laptop set up in my office at home and also do my “soft return” logon to allow my computer to do all the updates before I start back at work.

  • Please forgive me for not being a TWaT.

    Please forgive me for not being a TWaT.

    As we returned to presence working in July 2021, there was a major change from 2020. The previous summer, we were in in split shifts, meaning a week in the office, and then a week working from home in two fixed teams. The latter option meant less setting up and putting away of computers, but had the inflexibility that you only saw one half of your department. If there were two people working on a topic, they only ever saw each other virtually.

    (more…)
  • Back to the “normal” Office

    Back to the “normal” Office

    After having worked from home for sixteen weeks, last week I was back in my “normal” office, in the room that I have worked in since 2014, since I joined my employer.

    (more…)