About
With over 9 years of backend development experience—8 years specializing in Golang—I…
Activity
826 followers
-
Abrar Ahmed reposted thisAbrar Ahmed reposted this🌎 My "New Network" Ritual: Traceroute First, Connect Later. Whether it’s a hotel lobby, a corporate guest Wi-Fi, or retail free Wi-Fi, I have a ritual that happens before I do anything else. Before I toggle on my VPN or connect back to my home network via Tailscale, I’m pulling up a terminal to run a traceroute. This isn't just a geeky habit—it’s how I get a head start on knowing exactly how my VPN is going to perform. I need to see the "hops" and the raw path latency between my current location and my home servers before I introduce the overhead of a tunnel. Why the pre-VPN traceroute? • Performance Forecasting: If I see high jitter or packet loss on the third hop, I already know why my remote desktop session might lag later. • The "Northeast Loop": Living in the Boston area, I see this constantly. I’ll be trying to reach my home lab just a few miles away, but the traceroute reveals my traffic is taking a 400+ mile round trip down to a New York City and back just to cross town. • Peering Insights: I want to see how this specific ISP hands off traffic. Is the local network routing me through a weird gateway in another state? Once I’ve mapped the route and satisfied my curiosity, then I flip the switch, the VPN goes up, and I disappear into my private network. It’s about seeing the invisible infrastructure before I tunnel through it. 🔍 Real talk: Am I the only one who does this the moment they get a DHCP lease, or are there other network geeks out there who can’t help but learn the "shape" of the internet in their corner of the world? Let me know your go-to "new network" ritual in the comments! 👇 #Networking #HomeLab #Traceroute #Tailscale #VPN #NetworkEngineering #BostonTech #InternetInfrastructure #SysAdmin #Routing
-
Abrar Ahmed shared thisFinally here - AWS Community Day Lahore 2025 🎉 Ready for a full day of learning, inspiration, and builder energy. 🚀 Big thanks to the organizers and speakers for putting this together, excited to network, learn and absorb. 🙌 AWS Community Pakistan #AWSCommunityDay #GenerativeAI #CloudSecurity #Observability #BuildOnAWS #AWSPakistan #AWS
-
Abrar Ahmed shared thisI'm attending AWS Community Day Lahore 2025, let's meet in Lahore! As a Golang backend + video streaming engineer, I'm hyped for the agenda especially these sessions: - K8s on AWS - EKS in 20 Minutes - Monitoring Kubernetes Cluster on AWS (EKS) If you're working with Go, gRPC, or Kubernetes, let's meet up at the venue. #AWSCommunityDay #AWSPakistan #BuildOnAWS #CloudComputing #Kubernetes #EKS #AWS #awsusergroups #Networking #CommunityBuilders #LahoreEvents
-
Abrar Ahmed reposted thisAnother much‑deserved Emmy award in the streaming industry. This time, the Emmy goes to the CMAF standard - a technology that has significantly simplified the delivery of media in HLS and DASH formats. I’m most proud to be among the MPEG Systems Group experts and contributors to this standard receiving the award.
-
Abrar Ahmed reposted thisAbrar Ahmed reposted thisOEMs aren't incentivized to fix HEVC, they’re incentivized to move away from it. You know something is off when the hardware blocks are present in silicon, the cost has already been paid, and yet OEMs still disable them in firmware. That is not typical optimization behavior. It suggests deeper structural issues. Those issues come from the ecosystem itself. Twelve years after its introduction, HEVC licensing remains fragmented, unpredictable, and difficult to plan around; especially for manufacturers shipping high-volume, low-margin devices. Given that uncertainty and the tight margins on entry-level hardware, it wouldn’t be surprising to see even more devices ship with HEVC hardware decoding disabled. And this creates a long-term limit: HEVC adoption can grow, but it cannot become ubiquitous. The fragmentation and unpredictability baked into its licensing model ensure that it will never replace H.264 as a universal baseline. It’s a catch-22: the royalty model meant to fund research and innovation in a future ubiquitous video format ends up preventing that ubiquity. For streamers, this means paying royalties for a format whose device coverage is increasingly unpredictable. That’s a difficult foundation for long-term strategy. Just to be clear, this post isn’t about blame; it’s about rational actors responding to a broken system. #VideoCompression #StreamingTech #MediaTechnology #HEVC #H264 #AV1 https://lnkd.in/eksvMdXpDell and HP disable hardware H.265 decoding on select PCs due to rising royalty costs — companies could save big on HEVC royalties, but at the expense of usersDell and HP disable hardware H.265 decoding on select PCs due to rising royalty costs — companies could save big on HEVC royalties, but at the expense of users
-
Abrar Ahmed reposted thisAbrar Ahmed reposted thisHave a row contention problem? Try slotted counters! This is a neat technique for spreading out increments across many rows to reduce contention. Consider a website like YouTube, where we need to track view counts for every video. It's convenient to store this as a column in our video table, but for popular videos there will be lots of queries simultaneously attempting increments on the same row! Contention == locking == slow! The fix? Spread the counting out across N rows in a dedicated counter table. In the example below N=3, but you can make N=100 or N=1000 depending on the load. Every increment chooses one of the N rows randomly to increment. When we need the full count for a video, sum all counts for rows with a specific v_id. This can be done ad-hoc or on a regular interval and stored in the video table. More details here: https://lnkd.in/gGv8tr94
-
Abrar Ahmed shared thisPrompt Theory 🤔 What if the AI characters didn't know they were AI? (Made with Google's VEO 3) credit: Hashem Al-Ghaili
-
Abrar Ahmed shared thisNow AI generated Audio+Video with recently announced Google Veo 3Abrar Ahmed shared thisIs this AI or real? That’s the question now. When Sora first dropped, we all worried... deepfakes, propaganda, fake content... We knew AI video would get better. But this fast? 15 months later, Google Veo 3 is here and it’s honestly scary. Lighting, depth, motion… everything looks real. Other than some awkward dialogue, you can’t even tell. If this is where we are today… Imagine what the next 15 months will look like. Scary and impressive at the same time. credit : Hashem Al-Ghaili
-
Abrar Ahmed reposted thisAbrar Ahmed reposted thisThis is how millions of jeans get recycled into new pairs ♻️👖👖♻️ #recycling #recycle #jeans #landfills #Recycling #Sustainability #ReduceReuseRecycle #ZeroWaste #EcoFriendly #RecycleRight #CircularEconomy #GreenLiving #WasteReduction #RecyclingMatters #SDG #SDGs #ESG #sustainability #sustainable #ecofriendly #sustainableliving #environment #zerowaste #carbon #climatechange #nature #gogreen #recycle #savetheplanet #plasticfree #eco #reuse #design #greenenergy #architecture #innovation #renewableenergy #earth #climateaction #recycling #circulareconomy #climatecrisis #fossilfuel #pollution #CO2 #emission #agriculture
-
Abrar Ahmed liked thisAbrar Ahmed liked thisCPUs are getting worse. We’ve pushed the silicon so hard that silent data corruptions (SDCs) are no longer a theoretical problem. Mercurial Cores are terrifying because they don’t hard-fail; they produce rare, but *incorrect* computations! — *When* exactly the problem occurred is hard to pinpoint. The possibility was brought up at the Dependable Systems and Networks conference in 2008. The first real SDC disclosure happened in 2021 with Meta. Google and Alibaba also confirmed later. — Perhaps more terrifying is that cores can *become* mercurial over time. Chips are pushed so hard that electromigration aging can make compute “more wrong”. No one knows for sure what process node started the phenomenon...but it's statically likely to be 14nm or 7nm. — For decades SDCs were considered a myth, only caused by cosmic rays and such. We’ve only just entered the magic era where: 1. silicon is being pushed HARD 2. hyperscalers are getting SO big the issue is statistically visible Humanity is *just* starting to enter an era of needing to accept potentially imperfect compute. Current metrics imply about ~1 in 1000 CPUs are mercurial. Every indicator is pointing towards the issue getting worse in the future.
-
Abrar Ahmed reacted on thisThere is one more thing I’ll be doing early next year: teaching. It’s a classic graduate‑level course in digital image processing that has been part of MIT’s curriculum since the 1980s. The 2026 edition will cover many new advances in the field, including the growing use of machine learning techniques, HDR imaging, the latest image and video compression algorithms and standards, modern visual search and recognition pipelines, and more. The course is open to currently enrolled MIT students, but I plan to record the lectures and make the slides available for broader dissemination.
-
Abrar Ahmed liked thisAbrar Ahmed liked thisApparently, the map[T]struct{} trick for implementing sets doesn't save memory in Go 1.24+ map[T]struct{} now uses exactly the same amount of memory as map[T]bool. The reason is that in pre-1.24 maps, keys and values were stored separately in each bucket: first the keys, then the values: [K1, K2... K8] [V1, V2... V8] For the struct{} type, the value array occupied 0 bytes. In the new maps, keys and values are stored in a single structure (slot): [ {K1, V1}, {K2, V2}, ... ] struct { key keyType elem elemType } According to the compiler rules, if a struct ends with a zero-size field, that field still gets 1 byte of memory (in case someone tries to create a pointer to it). So, a field of type struct{} will take up 1 byte — just like a bool. The Go team is working on a fix.
-
Abrar Ahmed liked thisEhm, just because the current AI understands neither the problem asked nor its own output, the system never realized that it provided a proof? Aristotle (https://lnkd.in/eUabse76) "solved" also just the simpler version of the problem #124 and there are obvious signs that it just recited discussions between mathematicians found online (this would be a perfect match to my understanding of its "creativity" 😂 ). "After inputting the problem, it searched hundreds of millions of proof strategies and finally output a 100% verifiable theorem." The key word here and for #481 is *searched" millions proof strategies. " (quote: eu.kr36.com). Check another explanation in the comments. What are we celebrating here? A fast search engine? COOL! 🔥 What is happening? No human would have ever a chance to even read all the searched text in their lifetime. We waste energy of a small country to find summaries of **World-Wide** webcontent. The secret meaning of WWW in LLMs.Abrar Ahmed liked thisAI is ushering in a golden age of mathematics research. One of the greatest living mathematicians Terence Tao recently used Gemini DeepResearch for a literature review and noticed it had off-handedly solved Erdős problem #481 as part of its reasoning. It then moved on to explain why the problem is hard without realizing it had basically closed it. That’s the new pattern for mathematics AI: systems that can bulk-scan the landscape, synthesize prior work, and occasionally push all the way through to a real proof. And this happens while humans still provide taste, direction, and interpretation. The frontier shifts from “can we solve this?” to “which questions should we even be asking, and how do we turn a firehose of machine proofs into new theory?”
Experience & Education
-
GreenWire Solutions
View Abrar’s full experience
See their title, tenure and more.
or
Languages
-
English
Native or bilingual proficiency
-
Urdu
Native or bilingual proficiency
Other similar profiles
Explore more posts
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top contentOthers named Abrar Ahmed in United States
113 others named Abrar Ahmed in United States are on LinkedIn
See others named Abrar Ahmed