Jelena Kolomijec

Here's my LinkedIn post: **The Multiplatform Misfire 💡** Last month we shipped our Kotlin Multiplatform-powered mobile app and ran into some unexpected issues with code sharing between iOS and Android. Our devs spent hours debugging, only to discover that a seemingly innocuous API call was causing the crashes. **The Challenge:** Code sharing is one thing, but ensuring compatibility across different platforms is another beast altogether. We underestimated the complexity of mapping Kotlin's standard library to each platform's equivalent APIs. **The Lesson:** Don't assume that code will magically work on multiple platforms without proper testing and profiling. What works beautifully on Android might throw a tantrum on iOS (or vice versa). **Actionable Recommendations:** • **Test, test, test**: Create a comprehensive suite of tests for your Multiplatform project to catch issues early. • **Profile, profile, profile**: Use tools like LeakCanary or MemoryMonitor to identify performance bottlenecks and memory leaks. • **Keep platform-specific code separate**: Don't mix iOS and Android code in the same module; keep them separate to avoid headaches down the line. #KotlinMultiplatform #MobileAppDevelopment #CrossPlatformDev

1

1 Comment

I put together a write-up on write authority in distributed systems, mainly around stale writers, fencing, ownership transfer, and why retries can often lead to breaking correctness. Interested in feedback if you have any https://lnkd.in/gXFH_Sgz

10

2 Comments

"Research from job search company Adzuna showed a fall of 31.9 per cent in entry-level white-collar positions since the launch of ChatGPT in November 2022. Such positions have also fallen as a proportion of all vacancies. This confirms a wealth of anecdotes, and all of the signs are that this process will accelerate in the coming years, with one company after another replacing entry-level human employees with artificial intelligence (AI) applications and algorithms. BT, for example, has already announced 10,000 posts being axed in the next few years with the promise of more reductions to come. At present, the fall is in entry-level positions, both graduate and non-graduate and including degree apprenticeships, but this will spread rapidly to more established and higher-level positions. Government policy, such as the rise in employment costs due to employers’ National Insurance (NI) contributions, exacerbates this, but it is technology that is driving it." https://lnkd.in/eBz7WNHa

2

Kapil parekh is a rare one. XLR8 is already helping us manage our distributed services. We have no guarantee of the frequency of the time series data we're operating on (it can easily vary by 1000x). This tool helps us maintain our memory usage, while increasing query speeds by up to 4x; this reduces our operational costs while ensuring no OOM events.

6

1 Comment

The postal service debate shows how much profit obsession has rotted our brains. The headline "quarterly loss" implies that the correct headline is "quarterly profit", but what does that really mean? Here's some headlines that are aparently desirable if we assume profit is good: - Postal service charges more for providing essential service than it costs to provide - Savvy new CEO leverages government-granted monopoly to extract record profits while cost to send rises to all time high - Postal service cuts rural delivery routes and weekend service; reports record surplus If a public service isn't losing money, it's not a public service, it's the government taking your money, and we all know how much people love that.

39

9 Comments

I had my first piece published on HackerNoon. We often talk about tech disrupting industries, but who is it actually disrupting them for? The UK has introduced some of the strongest renter protections in a generation to tackle dangerous living conditions like damp and mould. Yet, the tech ecosystem remains starkly one-sided. We have endless tools helping landlords optimize portfolios, but almost nothing helping tenants seamlessly document evidence, automate compliance notices, or enforce their statutory rights. In this piece, I look at the massive gap between legislative intent and technical execution, why "Tenant Tech" is a massive, overlooked space ripe for meaningful disruption and why does PropTech always default to building for the landlord instead of the tenant? If you rent, work in tech, or care about how software impacts society, please give it a read and let me know your thoughts! 👇 https://lnkd.in/eP65rdEt #ProductManagement #Founders #SocialImpact #UKHousing #BuildingInPublic #CivicTech #Proptech #SoftwareEngineering

1

1 Comment

I didn’t get picked for the Lovable SheBuilds hackathon (bit bummed tbh), but I still want to share the two ideas I pitched because I think they’re worth building. CareShift → A simple way for unpaid carers in the UK to coordinate shifts, spot coverage gaps, and check Carer’s Allowance without trawling government sites. With 5.8m+ unpaid carers, it feels like a tool that’s badly needed. GrantPilot → A dashboard that pulls together UK grants and tenders (Innovate UK, councils, local funding). Right now, women-led businesses receive only a tiny share of grants and investment. Too many founders don’t even hear about the opportunities, or waste hours trying to navigate multiple sites. GrantPilot would match SMEs - especially women-led ones - with relevant funding, send deadline alerts, and provide draft application templates so they can compete on equal footing. Even though the hackathon didn’t happen for me, I’m still keen to build one of these. 👉 Which one do you think makes the bigger impact - CareShift or GrantPilot? 👉 Have you faced either of these challenges yourself? Would love to hear your thoughts.

14

4 Comments

I was thinking about some threads on LinkedIn about the place for leetcode style interviews. I think these sort of interviews are optimised for people under 30 who are still inwardly focused and can prioritise several hours a day to drill patterns. If I think about the persona of a typical 40 year old (extrapolating from myself) they are probably spending there evenings: - spending 3 hours before work and 3 hours after work managing their kids. - filling out insurance paperwork for shortfall from their kids last stay in hospital - possibly helping an elderly frail parent with their banking, or worse dealing with their deceased estate - and for good measure maybe dealing with a random audit from the tax man for a minor claim 4.5 years ago. For this persona there really isn’t time to spend drilling leetcode, and this style of interview is probably going to filter out people who are deeply experienced having 20 years of work behind them, and have mastered juggling life and work already.

10

1 Comment

Saw this article during work earlier and was like wait. This was published today?! After reading it I had a throwback moment to when I was trying an early version of Angular out in the wild. From what I recall, the concept was basically why load entire templates and pages from servers when the browser could handle this on its own? Well, now it just about can actually.

2

building ai features that work in demos is easy building ones that work consistently in production is harder next thursday, my superstar colleague Tynan Byrne reveals how we ship enterprise grade ai features to customers like spotify, monzo at Omnea come through to learn how to iterate methodically and know whether changes are genuine improvements Thursday 5th February - 6pm Wave Talent HQ - Old Street Sign up here 👉 https://lnkd.in/eGXhs2Us

12

As a dev watching social platforms evolve, I'm struck by an unexpected metric: X lost 10% of EU users while Bluesky's daily posts decline. The technical implications? Platform architecture choices now intersect with political identity in fascinating ways. It's not just about scalability anymore. What I'm observing: 🔹 Bluesky's decentralized protocol vs X's centralized model 🔹 User migration patterns affecting server loads 🔹 Content distribution challenges across fragmented networks 🔹 Authentication and identity verification complexities The most intriguing part? These technical decisions now carry political weight. Each architectural choice shapes community dynamics. Think about it: •Federation vs centralization •Content moderation capabilities •API access and developer ecosystem •Data portability options As engineers, we need to consider how our technical decisions influence platform demographics. These choices ripple through user behavior, engagement metrics, and ultimately, platform sustainability. What technical features do you think will shape the next generation of social platforms? How can we build for both scalability AND healthy discourse? #SoftwareEngineering #SocialMedia #TechTrends 𝐒𝐨𝐮𝐫𝐜𝐞:https://lnkd.in/gn_Fv4nK

5

The real competitive advantage is: Powered by Knowledge Driven by Passion Quote: “…. The chatty parrot of generative AI is no more than a linguistic trickster that is too flaky to be used to automate serious business functions. Lloyds has even begun offering insurance for “unforeseen performance issues” arising from “AI-driven products and operations”

1

On Thursday, during Droidcon London, I spoke in the Tech Leadership track about a simple truth: the hardest problems in software engineering are rarely about code. They’re about people, structure, and habits. AI won't change it, but could help. Across many teams, I’ve seen the same dysfunctions appear: unclear ownership, review processes that create friction instead of alignment, ego, etc. These patterns aren’t isolated. They repeat because most teams focus on fixing symptoms instead of addressing the system. Real progress comes from small, continuous adjustments that strengthen culture and accountability. When you make improvement part of the workflow instead of a side project, velocity and quality start to reinforce each other. The discussions after the talk made one thing clear: these challenges are universal, but so are the solutions. Slides are available here:

29

1 Comment

Very (very) niche tool but if you're interested in analysing different industries in the UK, each company has a code called a SIC code, and these codes are actually hierarchical, which means that code 64301 and 64302 are in; - the same division (64, which is financial service activities) - the same group (3, which is trusts, funds, and similar entities) But differ only that 64301 is investment trusts and 64302 is unit trusts. This site helps break down what the codes mean https://lnkd.in/edRbBpus This means that if you have the SIC codes for a bunch of companies, you don't need to do any further processing/labeling, since the labeling is baked in to the codes. So to analyse e.g. all financial services companies, you just match code 64* (but just note that some divisions/groups are really broad, so that type of grouping isn't always meaningful).

4

2 Comments

Speaking at APIdays London reminded me why we built APIs in the first place, to connect systems seamlessly. But somewhere along the way, REST endpoints multiplied into coordination nightmares. My talk explored how GraphQL brings us back to that original vision: one interface, infinite possibilities. The conversations afterward convinced me we're at a tipping point where APIs either become innovation accelerators or development bottlenecks. Thanks to the APIdays team for organizing an outstanding event, and to the engaged audience who proved this resonates across the industry. #APIdays #GraphQL #APIDevelopment

27

3 Comments

A founder recently built an AI voice agent named Rachel to solve a simple but persistent problem. After noticing high prices for a pint of Guinness in Dublin and learning that official tracking of pub prices had stopped years ago, he had the agent call more than three thousand pubs across Ireland. The goal was straightforward: gather real time data on Guinness prices to create greater market transparency. The results were notable. Pubs began adjusting their pricing once the compiled information started circulating, showing how accessible data can quickly influence competition and consumer choice. On one hand, this demonstrates the practical power of current AI tools. Voice generation combined with conversational models allowed one person to collect information at a scale that would have been impractical manually. It highlights how AI can fill information gaps and drive more efficient markets. At the same time, the approach raises valid questions about consent and transparency. Many bartenders had no idea they were speaking with an AI, and some reacted with frustration at receiving unsolicited calls. When deployed at scale, such methods risk being perceived as intrusive or even crossing into unwanted automation that wastes staff time. As organizations explore similar agentic applications, the recommended path is to prioritize clear disclosure and ethical design from the outset. Build systems that identify themselves early in the conversation and offer value in return, whether through aggregated insights shared publicly or opt in participation. This maintains trust while still capturing the efficiency gains. The real test for AI adoption in business will not be technical capability alone but whether implementations respect the human context in which they operate. Overlooking that distinction can turn a clever solution into a source of backlash. What boundaries should companies establish when using autonomous AI agents for data collection or customer outreach? To Know more https://lnkd.in/ddr-gCb7 #ArtificialIntelligence #BusinessStrategy #EthicsInTech #MarketTransparency

3

Tuned in to Dom Davis AI: Assisted Ignorance talk last night at Dick's Bar, Norwich. Thank you for the talk, Dom! It was great to get outside of my usual echo chamber and listen to first hand experiences of people’s views on AI. Some points that resonated with me: * AI may increase employee workload rather than decrease by devoting time to wrangling and reviewing it's outputs. The key insight being AI is not giving the brain the break it thinks its getting and the body can’t take this over long periods of time. We have absorbed the narrative that AI = productivity, but maybe we need to slow down and re-fine our own workflows first. * We are still yet to see a winner in the ‘AI-wars’ and it’s important to not get locked into one stack - I have used a handful of different LLMs and agentic AI approaches so far and the speed of innovation is remarkable. My own take, we gradually see more interest and investment in building and maintaining specialised ‘local’ models in local data centres to seamlessly answer the question - ‘which bin do I need to put out today?’

5

The Queue on Tempo (ADHD Task Master) The Queue (seems like a bad ass London underground club, thats obnoxiously hard to get in) is my fancy word for a task backlog but its actually more simple than conventional productivity apps. KEEP AN EXHAUSTIVE LIST OF ALL THE THINGS YOU NEED TO DO You determine this list of things--just a list of things, actions, chores, work related and projects-- just work! You can create this list anywhere: by using the "Create Tasks" tool or descoping incomplete tasks from a current "Sessions". FROGS + PRIORITY SORTING You determine how important this thing is to you and how much you hate doing it (we call this a "frog"). Frogs you need to eat first. You need to rip off the band aid. These frogs manifest themselves in mental rot or menial tasks but they are great for you to complete, because really they lift such a weight off of you to get you through your day. Say its doing your least favorite subject (Calc?) ... You can even create reoccuring frogs, that you can slot into your day but not on a date specific/calendar/alarm. You know this task needs to be done (folding f*cking landury), so as you create a session, you can pull it in! CREATING SESSIONS FROM THE QUEUE The queue also allows you to timebox the session you want to try and hammer out. They have quick options (I've got two hours) and it takes the time you estimated for each task and pulls the suggested tasks in. IMPORT There are many ways to build tasks and sessions, one way is just a simple list you might have wrote on a digital notepad or doc. You can pull that over. And voila, a way to keep a tab on future things and hammer at them a day at a time :)

7

2 Comments

When I joined zally®, almost everyone asked me the same question: "So... what does zally actually do?" It felt like I'd joined some kind of mythical creature in the Manchester startup scene. Everyone had heard of zally, and they had seen the bold branding, but nobody could quite explain it. zally® analyses how you use your device. Your typing rhythm. The way you touch the screen. How you hold your phone. Passively, in the background, with no prompts or interruptions - it builds continuous confidence that the right person is behind the screen. No passwords. No friction. Just quiet, intelligent security. We've been building this for a while, and now it's ready for the world to see. The team at zally is the best and most dedicated I've ever worked with - and perhaps that's exactly what you'd expect from the team defining a new field of AI. We're opening up a small pilot programme - inviting a select number of organisations to integrate our continuous authentication technology directly into their iOS and Android apps. If your team is thinking about security posture, fraud prevention, or just making life easier for your users, I'd love for you to get in touch https://lnkd.in/eMEbb2K8 ✨

33

4 Comments