Stanislav Vishnevskiy

*Outsource Optimization Coding* After a three-year hiatus, I’m thrilled to be jumping back into coding - with a talented team alongside me! 🚀 Experts at fine-tuning games to run perfectly, particularly with CPU and memory usage optimizations. At my previous companies, I was always the “first contractor” from which we grew our contracts with clients, expanding each team from “just me” to 35 (Coconut Lizard) and to 45 (Pitbull Studio) people… only hiring the very, very best people - those passionate and self-driven who don’t need micro-managing. If your game is stuttering and not managing to maintain a solid 60-120fps, we can dive in and help you get there - whether by optimising code ourselves or by letting your artists, designers and musicians where they need to focus their efforts. In my own case, you can see some of my successes with Wheelman (a game not at all suited for Unreal Engine 3 - but which we got running at full framerate on Xbox 360), Unreal Engine (my studio were contractors to Epic during development - and later became Epic Games UK), Gears of War, Fortnite, Conan Exiles, Lawbreakers and more. My team have worked on similarly impressive games - and even more so, actually. Now, as I make my return, we're also looking to grow our available team. Whether you're a seasoned C++ guru or a performance optimization enthusiast, there might be a spot for you. We're tackling new challenges and ready to extend our collaboration to include fresh, driven talent. Interested in making games run smoother and faster? Please reach out. It can be remote or, better still, join us in Thailand! Keen to chat about potential collaborations or opportunities? Drop me a message here. Let’s get to know each other (if we don’t already) and make some magic happen! Can't wait to see what we can achieve together in this next chapter!

70

7 Comments

Vizzly's changelog for May: 🔌 Plugin API / Custom Views • You can now inject custom views from third-party libraries into the Vizzly dashboard. 🇺🇸 New US region • We launched a new server in the US (us-east-2). Performance will always be a priority at Vizzly; we know there's nothing worse than waiting 419 years for a dashboard to load. ➗ New Custom Metrics available • Users can now calculate the difference between two date values and create new metrics based on conditions set before aggregation. 📆 Relative Filtering • When applying dashboard-level date filters, you can now set relative dates. For example, you can filter data for 4 months in the future or past. ✨ Other developments • ID Aliases • Value Aliasing • Redshift 'super' data type support • Snowflake array support ✌️

38

1 Comment

I could sell you our platform code without remorse. Still, it won't work for you in 99% of cases. Why? Code is just a piece of writing. For the operator, it holds no value without context. A few software engineers might disagree, but in the grand scheme, it’s true. To understand the code, you, the buyer, need a CTO and a team of engineers. It might take a year to fully analyze it. During this time, you have to pay salaries for a team that might conclude: "The code isn't perfect. We need more time to fix bugs and add features to make it market-ready." By the time the analysis is complete, the technology will be a year behind. Meanwhile, the GR8 Platform keeps improving while your code stays the same. In a year, the platform has new integrations, improved features, and better performance due to optimization. Believe me, the real value of a platform lies in its product and engineering expertise, built on years of research and experience. You cannot go around it. So, even if I had the opportunity and would agree to sell, it's not me who'll regret this decision in the end. #GR8Tech #igaming #sportsbook #igamingplatform #bettingprovider #iGamingSoftware #softwarearchitecture

75

8 Comments

IMO one of the most underrated Guardrails AI features is streaming support. E.g. here's streaming in real time to show how you can validate that the LLM output doesn't contain PII. The difference in latency is wild! Code snippet in comments below:

219

13 Comments

🚨 ChatGPT4 and Live Ops/Gaming 🚨 With ChatGPT4 here to stay and all the advancements google announced yesterday with Gemini at I/O, many industries are excited about the potential impact of this ever evolving technology as I was back in 2011 with IBM Watson opening up as an API for developers. In the mobile gaming industry, ChatGPT 4 could revolutionize the live ops model, creating more immersive and personalized gaming experiences, and real-time support for players. But the concept is easier said than done and the business model yet not scalable to most gaming studios. One of the pros of ChatGPT 4 is its ability to enhance player engagement. With dynamic dialogue, players can enjoy more personalized gaming experiences, keeping them engaged for longer periods. Additionally, the natural language processing capabilities of ChatGPT 4 can offer real-time assistance to players, improving overall player satisfaction and breaking global barriers of idioms, literacy and accessibility. Another advantage is the ability to generate dynamic in-game content tailored to individual player preferences. This feature can lead to higher retention rates, which is crucial in the highly competitive gaming industry. However, there are also a few cons to consider. For instance, integrating ChatGPT 4 into existing live ops models may require significant development resources and expertise, posing a challenge for smaller game studios. Utilizing a chatbot with access to player data also raises privacy and security concerns, necessitating robust data protection measures to mitigate risks. Finally, the initial investment required to license and deploy ChatGPT 4 at scale could be prohibitive for some companies, especially smaller studios with limited budgets. Despite these challenges, the transformative potential of ChatGPT 4 in revolutionizing the mobile gaming industry's live ops model cannot be overstated. It's a learning journey, and I invite you to share your thoughts and POV on how developers can leverage this technology to create even more engaging and immersive gaming experiences! What's your opinion ?

16

2 Comments

Put some thoughts together on why we think open source works for SSOReady. I was convinced to edit out my convoluted LTV model, so hopefully it's not nearly as dense as some of my other blog posts :) https://lnkd.in/g_mXsprU

30

Agreed! That's why we've been expanding our 3D asset management capabilites at VNTANA. We completely agree devs should stop wasting time which is why we have: -Intelligent Optimization to automatically reformat and resize 3D assets -Robust search and meta data tied to assets -Approval workflows for more automated QA -Versioning & user change logs coming very soon And agree the company that can becomes the repository of the majority of 3D assets is well suited to win the 3D AI space.

23

Custom notifications and advanced filtering are here!

7

Got an email "GCP's Apache Kafka service is now live" at noon. After dinner, spent half an hour confirming it works well with Timeplus. I don't think you can make the bootstrap server publicly accessible. Has to setup a bastion host or run your app within the same project/VPC. Setting up the SASL username/pwd was not easy. Follow the guide carefully https://lnkd.in/gcKEVh3g and you may need to disable the iam.disableServiceAccountKeyCreation Anyway, if your workload are in GCP already, having such Kafka service in GCP is nice way to burn your cloud credits. Yet another Kafka vendor, great.

33

2 Comments

Self organizing files, yet another l cool innovative use of LLM from Llama team to reorganizing, renaming your files, bring sanity to disorganized files and clutter. Watch the video on the link below https://lnkd.in/gteHHZ_D Video: https://lnkd.in/gWKpjTJi #llm #innovation #llama #ai #operatingsystem

18

5 Comments

🤔 Creating good prompts can feel like a head scratcher, right? We're releasing a FREE tool to generate high quality prompts for Claude 3 Opus 👇 Our free tool, powered by Anthropic's metaprompt, helps you become an advanced prompt engineer for Claude 3. No prior experience, technical knowledge or API key required To use it just describe your task in a few words, and the tool will generate a Claude Opus prompt for that task. I haven't come across an easier tool than this to generate great prompts. Don't believe me? Try it yourself: https://lnkd.in/gBGcB7JX Some example tasks for inspiration: - Choose an item from a menu, given user preferences - Rate a resume according to a rubric - Explain a complex scientific concept in simple terms - Draft an email responding to a customer complaint - Design a marketing strategy for launching a new product

53

8 Comments

🔥 People asked, and I've delivered: you can now make streaming data generators that dynamically adjust depending on the time/day. There are heaps of use cases for this: - spike latency by 15x during all North American holidays - burst the number stock trades for 5 minutes on market open/close - generate 1/200th as many retail transactions outside normal business hours Instead of fussing with time zones and off-by-one errors, I made that all my problem and tucked it behind a declarative interface. In the video, I show how a latency value usually hovers around 10, but every other minute it bursts to 50, and every 5th minute it bursts to 150. All in 8 lines of code. ❤️ Huge thanks to my customers who pushed me to build this.

69

8 Comments

We've just shipped webhook verification support for Tebex (Overwolf), a platform for game monetization. We support common webhook signature verification mechanisms out of the box but spend the time adding direct support for platforms because: 1. It improves the DX of using Hookdeck 2. Many platforms do webhooks signing differently (and that won't change any time soon) Please give the team a shout if you see a webhook provider you would like added to our directly built-in list (link in comments).

11

1 Comment

Version 2.2.7 of our Perfmatters plugin is live! - *New* perfmatters_used_css_below filter. 👨‍💻 - *New* perfmatters_defer_js_exclusions filter. - *New* perfmatters_delay_js_fastclick filter. - Added additional DOMDocument flag to parent exclusion filters for better compatibility. 🔧 - Added GiveWP request parameter to excluded page builders array. - Updated delay JS quick exclusion for WooCommerce single product gallery with additional scripts to help with zoom + lightbox functionality. 🖼️ - Moved Script Manager CSS to stylesheet printed inline instead of from a PHP file. - Updated delay JS quick exclusion for Cookie Notice with additional scripts. 🍪 https://lnkd.in/gy_U_Rk #WordPress #webperf

17

2 Comments

ELO Ratings for flyflow-voice-small simulated on real-world conversations with a GPT-4 evaluator. Barely edges out GPT-4 on ELO and for latency (lower is better), we're on average 4.2x faster to respond.

4

GPT-4-Turbo 🆚 GPT-4o Mikhail Burtsev recently shared some interesting findings on the performance of different GPT-4 models. According to his tweet from ICLR 2024, GPT-4 Turbo shows impressive results for simple question answering over documents up to 16K, while GPT-4o excels in shorter contexts up to 2K due to its lower cost and faster speed. For those dealing with mid-sized documents (6-10 pages), GPT-4 Turbo could be the way to go, offering higher quality outputs. On the flip side, the modifications in GPT-4o, aimed at increasing its speed, present a tradeoff that should be considered based on your specific needs. Before choosing your GPT-4 model, take a moment to consider which one aligns best with your tasks. 💡 Link to the original tweet by Mikhail Burtsev: https://lnkd.in/eEJ8695g #AI #GPT4 #MachineLearning #TechInsights #ICLR2024

3