West Bloomfield Township, Michigan, United States
34K followers
500+ connections
34K followers
500+ connections
Articles by Sidney
Activity
34K followers
-
Sidney Jones shared thisCan your AI do this? Over the past 18 months, I’ve gone deep into AI research, and that journey completely changed how I build. We’re now creating powerful, high-fidelity apps in about 30 minutes. This finance demo is one example. The breakthrough isn’t just speed. It’s architecture and control. Most teams still use AI in ways that create accidental complexity: - Low-fidelity outputs - Non-deterministic behavior - Context contamination - Hallucinated implementation details The coding constraints may have dropped, but the bottleneck moved to architecture. My focus: - Converting intent into constraints - Defining guardrails that produce deterministic outcomes - Raising agent benchmarks so AI builds what we mean, not what it guesses This video series starts with: Can your AI do this? Then we unpack exactly what our AI is doing for us, and how we can help you uplevel your own workflow to do the same kind of magic. If you want warp-speed delivery without losing system integrity, this is for you. #AI #SoftwareArchitecture #AgenticAI #ProductEngineering #EnterpriseAI #Automation
-
Sidney Jones posted thisIn leadership, I’ve learned that feeling the pain points is part of the job. Often, you end up experiencing the very friction you’re there to solve: ✅ misalignment ✅ unclear expectations ✅ communication gaps When I reflect on this more deeply, I’ve started to see “pain” the way the body uses it — 🛜 as a signal. As an engineer, signals matter. They help diagnose root causes instead of just treating symptoms. In leadership, I try to take the same approach. I ask for feedback. I don’t like silent failures. I want honesty — the reality of how I’m being experienced. When I ask for feedback — especially when I think things are going fine — I’m looking for those signals. How am I being perceived? Where is friction forming? What am I not seeing? Because better diagnosis leads to better solutions — implemented with clarity and compassion. #leadership
-
Sidney Jones posted thisFrom “Code as Documentation” → “Code is Documentation” Let’s unpack why this matters so much. The Old Idea: Code as documentation This phrase always meant: • “We failed to write good docs” • “Just read the code” • “The truth is implicit if you’re smart enough” Which leads to: • tribal knowledge • gatekeeping • accidental complexity • docs that rot • onboarding hell It’s defensive. It assumes humans must adapt to machines. The New Reality: Code is documentation This flips the direction entirely. Now it means: • intent is explicit • behavior is declared, not inferred • structure is readable • evolution is traceable • meaning survives refactors It assumes machines must respect human meaning. That’s a philosophical inversion. Most systems cannot make “code is documentation” real because: • code is opaque • behavior emerges implicitly • documentation is bolted on later The “code”: • is written in human language • is version-controlled • is reviewed like prose • is executable through projection • is the only source of truth There is no secondary explanation layer. So yes - the code is the documentation. ________________________________________ This Quietly Solves a Huge Trust Problem When code is documentation: • auditors can read it • teachers can explain it • AI can reason over it • new developers can understand it • users can trust it No more: “The docs say X, but the system does Y.” Because there is no separation. #loga #thesourceoftruth
-
Sidney Jones posted thisLet’s face it, the tech industry is evolving too fast for degree programs to keep up. By the time a curriculum is approved, taught, and graduated, the tools, architectures, and best practices it was designed around are already outdated. That doesn’t mean education is broken—but it does mean learning has shifted. The highest-leverage skills today aren’t memorizing frameworks or languages. They’re: • Learning how to learn • Pattern recognition across domains • Systems thinking • Adaptability in real-world problem spaces • Building, shipping, and iterating in public The people pulling ahead aren’t waiting for permission from a syllabus. They’re learning from: • real projects • open-source ecosystems • mentors and practitioners • experimentation and feedback loops Degrees still have value—but velocity now matters more than credentials. The future belongs to those who can continuously retool themselves while the game is changing mid-play. And that’s a very different kind of education. #education #evolution
-
Sidney Jones shared thisDeploy $20B directly into massive GPU infrastructure; Nvidia-aligned priority access converts capital into a compounding, time-based competitive moat. The real AI moat isn’t the model. "It’s the clock. ⏱️" Everyone is talking about the $20B headline. But the "strategic advantage" isn’t the capital itself — it’s "how fast that capital converts into irreversible infrastructure". This diagram shows what I call the "Compute Infrastructure Flywheel". As an infrastructure architect, the goal is simple but brutal: > "Build compute capacity faster than competitors can fundraise and deploy it." Why? Because in AI, "time cannot be bought back". When $20B is deployed directly into massive GPU clusters — "with priority access through Nvidia alignment" — something critical happens: • 100,000+ GPUs arrive "while others are still in line" • Data centers spin up during a global supply constraint • Training runs scale immediately, not “next year” • Better models emerge sooner • Revenue arrives earlier • Capital gets reinvested before competitors even start That creates a "12–24 month time-based moat" — and time-based moats "compound". Even if a competitor raises the same amount tomorrow, they’re still waiting: • for GPUs • for power • for networking • for cooling • for deployment • for iteration Capital is abundant. "Infrastructure lead time is not." This is why AI competition increasingly looks less like software and more like "industrial-scale orchestration": > "More infrastructure → Better models → More revenue → More infrastructure" Once this flywheel is spinning, it’s no longer about who has the smartest researchers — it’s about "who can sustain velocity while others stall". This is how compute becomes strategy. This is how time becomes a moat. (Sequence adapted from the xAI infrastructure orchestration analysis ) #AIInfrastructure #ComputeMoat #SystemsThinking #Architecture #CapitalAllocation #TimeBasedMoats #AIStrategy
-
Sidney Jones shared thisMost of our regrets don’t come from what we feel — they come from how fast we act. This article explores the moment where freedom actually lives: the pause between trigger and reaction. When we cultivate awareness, we insert choice where there was only momentum. That single skill changes conversations, leadership, relationships, and life itself.Cultivating Your Awareness Operating System: A Practical GuideCultivating Your Awareness Operating System: A Practical Guide
-
Sidney Jones shared thisWhy Chevron jumped on Venezuela headlines isn’t about “oil prices” — it’s about positioning. I mapped the Venezuela–Chevron story as a simple sequence: Policy → Capital → Production → Market Repricing What the diagram makes clear: • Policy shifts don’t move barrels immediately • They move **optionality** first • Markets price **who is already in position**, not who *might* enter later Chevron is unique here. While others would need: licensing → negotiations → JV formation → infrastructure rehab Chevron already has: • an operating footprint • JV relationships • waiver familiarity • execution muscle That’s why the market reacts *before* production changes. Exports can reroute faster than fields can recover. Infrastructure takes time. But "expectations move instantly". This is a classic example of how markets price: "control points in a system", not just outcomes. The diagram below shows the full chain: policy signal → sanctions path → capex → services → refining → investor repricing. If you’re analyzing energy, geopolitics, or any regulated industry, this pattern shows up everywhere. Markets don’t wait for certainty. They wait for "positioned advantage". #EnergyMarkets #OilAndGas #Chevron #Geopolitics #SystemsThinking #MarketStructure #Investing #Macro
-
Sidney Jones shared thisMost companies think they’re “data-driven.” Very few are actually closed-loop systems. This diagram captures the difference 👇 Sense → Decide → Act → Learn With the missing piece, most organizations forget: 👉 returning impact back into the market. What separates elite operators (in housing, software, finance, or any complex domain) isn’t prediction accuracy. It’s orchestration under uncertainty. 🔹 Sense Markets shift continuously: rates, demand, costs, behavior. 🔹 Decide Not with gut instinct — but with pattern recognition + domain constraints. 🔹 Act Pricing, inventory, incentives, capital allocation, execution speed. 🔹 Affect the Market (this is the step many miss). Your actions change buyer behavior, competitor reactions, and supply dynamics. 🔹 Learn Feedback updates playbooks, priors, and future decisions. This turns an organization from a [reactive participant]→into→[an adaptive control system] Most companies only run internal feedback loops (KPIs → dashboards → meetings). The real advantage comes from mastering the external loop: Action → Market impact → Market response → Signal → Adapt → That’s orchestration. → That’s resilience. → That’s how systems survive cycles instead of being victims of them. This same pattern applies to: * housing * software platforms * AI agent systems * capital allocation * leadership itself If your system can’t feel the market reacting to its own move, it’s not adaptive yet. #SystemsThinking #Orchestration #Leadership #Architecture #AI #AdaptiveSystems #Strategy #ClosedLoop #PatternRecognition #DecisionMaking
-
Sidney Jones shared thisAI Agents are finally moving from theory to ROI. And one company is leading that transformation: AMESA. Founded by Kence Anderson — who previously built over 200 AI agents at Microsoft — AMESA is on a mission to make AI agents operate in real industrial environments and generate measurable ROI. By fusing lessons from enterprise AI with real-world operations, AMESA created a new category of applied intelligence — and it’s already valued at nearly $40 million. Until recently, many thought multi-agent systems were still years away from practical deployment. AMESA’s breakthroughs show that the future of agentic AI is already here. 💡 What Makes AMESA Different Multi-Agent Architecture — a framework for agents that act, learn, and deliver business outcomes. Feedback: Agents learn by trial and error through simulations. Optimization: Continuous improvement toward goals. Goal Orchestration: Orchestrators assign and manage agent roles. Simulation: Agents test and refine real-world solutions using data. Policy Agents: Guide orchestrator decisions with prompts + RAG. Communication Agents: Explain agent actions through clear data and documentation. What stands out most is AMESA’s mindset: “Deploy agents where failure isn’t an option.” That standard forces engineering excellence — and their systems thrive under it. 🧠 What They’ve Built 📌 AMESA Orchestration Studio A no-code + API command center to design, manage, and optimize teams of agents. Fortune 500 companies are already using it to build agent hierarchies with live feedback loops. 📌 AMESA Agent Cloud A secure proving ground where agents learn, test, and improve. Connect data, attach digital twins, and simulate complex environments for safe, feedback-driven learning. 📌 AMESA Assist Agents LLM-based assistants that explain actions, provide context, and support better decision-making. The takeaway: The real ROI of AI Agents doesn’t come from theoretical power — it comes from performance under pressure. AMESA proves that AgentOps has officially entered the real world. #AI #AgenticAI #AgentOps #AMESA #LLM #RAG #MultiAgentSystems #AIEngineering #IndustrialAI #Automation #AIArchitecture #Innovation #FinOps #RenderX
Sidney Jones commented on a post
4mo
Resilient
-
Sidney Jones liked thisSidney Jones liked thisControl the tone, control the conversation. The smartest response isn’t louder...it’s calmer. Because you don't win difficult conversations. You manage them. Emotional intelligence is knowing how to respond without escalating. Keep these 3 insights in mind... 1. These responses protect boundaries without triggering escalation. None of the statements directly attack the other person. Instead, they redirect behavior while maintaining composure. High emotional intelligence means correcting dynamics without activating defensiveness. 2. Language is used to shift the frame, not win the argument. Many of these responses subtly change the conversation from emotional to objective (facts, goals, next steps). The goal isn’t dominance...it’s regaining control of the interaction. 3. Emotional control is demonstrated through pacing, not wording. The effectiveness of these responses depends less on the words themselves and more on delivery (tone, timing, and calmness). The real signal of emotional intelligence is staying regulated while others are not. Now read them one more time with these insights in mind... 10 SMART RESPONSES TO DIFFICULT PEOPLE When they interrupt you: “I’ll finish my point, then I’d love to hear your thoughts.” When they dismiss your ideas: “That’s one approach. Here’s another way to look at it.” When they get aggressive or emotional: “I’m happy to continue this conversation when we can keep it productive.” When they refuse to engage logically: “What evidence supports that? Let’s look at the facts.” When they try to dominate the conversation: “Let’s make sure we’re both heard. I’d like to share my perspective too.” When they make it personal: “Let’s keep this about the issue, not each other.” When they keep pushing a losing argument: “It sounds like we see this differently. What’s the best next step?” When you need to shut it down professionally: “I value this discussion, but I don’t think we’re making progress. Let’s revisit later.” When they try to control the narrative: “Let’s refocus on the main goal here.” ----- ♻️ Like, follow, and repost if this resonates. ➕ Follow Travis Bradberry for more and sign up for my weekly newsletter at TravisBradberry.com Thanks to Haris Halkic for the graphic. Do you want more like this? 👇 📖 My #1 bestselling new book, "The New Emotional Intelligence" is now discounted on Amazon.
-
Sidney Jones liked thisSidney Jones liked thisGhost Kitchen (Part 6) — Refactoring for a "Closed-Loop" Architecture 🏗️🍳 This weekend was all about moving from a simple "alert and fix" script to a more stable, closed-loop system. I’ve been refactoring the architecture to make the automation more transparent—so it doesn't just act, but actually "thinks" and verifies its work. Here’s how the new iteration fits together: 🛰️ 1. Real-Time Radar (The Monitor): Using Apache Flink and Redpanda Data, the system acts as a high-speed radar. It's constantly watching the "kitchen" telemetry and flagging a performance breach (SLO) the second things stay slow for too long. 🧠 2. The Reasoning Loop (The Brain): This is the big addition this week. When an alert fires, a LangGraph Agent steps in using the ReAct pattern (Reason + Act). * The "Reason" bit: It doesn't guess. It uses Agentic RAG to "read" my technical runbooks (stored in ChromaDB) to find the exact solution—like "scale up the pizza station." * The "Act" bit: It explains its thought process in the dashboard before it touches the cluster. 🛠️ 3. Self-Healing (The Action): The Agent then talks directly to Kubernetes to add more resources (pods) exactly where the bottleneck is. 🔄 4. The Reflection (The Verification): This is the part I’m most excited about. The Agent doesn't just run a command and walk away. It stays in a Reflection loop, continuously watching the live data stream until it verifies the performance is back to normal. It won't close the incident until the situation is actually resolved. The Big Picture: I'm still learning and iterating, but the goal is to see how we can move away from manual firefighting and toward a system that can observe, decide, and verify its own fixes. Watch the demo here - https://lnkd.in/gSZxMQhE My 1st Kubernetes Cert of 2026 - https://lnkd.in/gcYP6KWP #SRE #PlatformEngineering #BuildingInPublic #Kubernetes #AIOps #Automation #LearningInPublic #SystemDesign #elliottArnold
-
Sidney Jones liked thisSidney Jones liked thisArchitecture decisions age fast. Their consequences don't. Six months later, nobody remembers why the system looks the way it does. The code shows the outcome. But the reasoning behind it is gone. I ran into this recently. A new engineer asked: “Why are we using Clean Architecture here instead of a simple N-Layer design?” The team had three different explanations. → testability → separation of concerns → team preference None were fully right. The real reason was controlling complexity as the system grew. With N-Layer we started seeing: → domain logic leaking into infrastructure → services tightly coupled to frameworks → refactoring is getting harder every quarter So we made a decision. We adopted Clean Architecture across services. Not just for technical reasons — but to align teams. When engineers move between services, the structure is familiar. One architecture. One shared mental model. The problem wasn't the decision. The problem was that the reasoning behind it disappeared. That's when we introduced Architecture Decision Records (ADR). An ADR captures: → context → options considered → decision → consequences Good and bad. They live next to the system documentation, so anyone can understand why something was built the way it was. When a decision ages poorly, we don't delete it. We supersede it with a new ADR explaining what changed. The impact was immediate. → faster onboarding → fewer repeated debates → visible architectural consequences And the biggest surprise? Writing an ADR forces you to think more clearly. Weak decisions often fall apart when you try to explain them. Good architects don't just make decisions. They make their decisions understandable. How does your team document architecture decisions? ADRs, RFCs, something else? #SoftwareArchitecture #ArchitectureDecisionRecords #CleanArchitecture #TechLeadership
-
Sidney Jones liked thisSidney Jones liked thisGoogle just dropped a free AI arsenal. And barely anyone noticed. While everyone's paying for ChatGPT Plus and Claude Pro, Google silently built an entire AI ecosystem. Completely free. Production-ready. Available now. Here are 13 Google AI tools you can start using today: 1. Gemini Live → Voice + screen sharing AI conversations → Real-time problem solving 2. NotebookLM → Any source → instant summaries → Cuts research time by 80% 3. Veo 3 → Text → high-quality video → No editing experience required 4. Gems in Gemini → Build task-specific AI assistants → Automate your workflow 5. Google AI Studio → Test and compare AI models side by side → Experiment without limits 6. Google App Builder → Prompt → working app → Zero coding skills needed 7. Firebase Studio → AI-powered app development → Ship faster than ever 8. Imagen → Text → production visuals → Seconds, not hours 9. Gemini Ask on YouTube → Extract answers from any video → Save 30 minutes per video 10. Gemini in Google Sheets → Auto-generate formulas and insights → 10x your spreadsheet speed 11. Google Cloud Vision API → OCR + image detection → Enterprise-grade intelligence 12. Nano Banana → AI image editing with variations → Refine visuals instantly 13. Google Classroom AI Tutor → Personalized learning assistant → Adaptive education at scale Google stopped being just a search company. It became an AI infrastructure layer most people sleep on. All free. All accessible. All production-ready. Want to stay ahead of AI? Subscribe to TechSparksAi. → 3,000+ AI prompts → 10+ paid courses (free) → 1,000+ exclusive resources → ChatGPT cheat sheets Credit to Darshal Jaitwar. Follow him for more. __________ 𝗕𝗲𝗰𝗼𝗺𝗲 𝗯𝗲𝘁𝘁𝗲𝗿 𝗮𝘁 𝗔𝗜 𝗶𝗻 𝗷𝘂𝘀𝘁 𝟭 𝗺𝗶𝗻𝘂𝘁𝗲 𝗮 𝗱𝗮𝘆. 𝗚𝗲𝘁 𝘁𝗵𝗲 𝗔𝗜 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿 𝘀𝗺𝗮𝗿𝘁 𝗹𝗲𝗮𝗱𝗲𝗿𝘀 𝗿𝗲𝗮𝗱. 𝗦𝗶𝗴𝗻 𝘂𝗽 𝗳𝗿𝗲𝗲 𝗻𝗼𝘄 → aiforleaders.com
-
Sidney Jones liked thisSidney Jones liked thisYour API works perfectly - until someone hammers it with 10,000 requests in a second. Rate limiting is what stands between a stable system and a full outage. But not all rate limiting algorithms are equal 👇 1. Fixed Window Counter Counts requests in a fixed time window and resets after each interval. Simple to implement but burst-prone at window boundaries. 2. Sliding Window Log Stores each request timestamp and removes expired entries. Accurate limiting but memory-heavy at scale. 4. Sliding Window Counter Combines current and previous window counts to smooth traffic. Lower memory usage, better burst protection than fixed windows. 4. Token Bucket Adds tokens at a fixed rate. Requests consume tokens. Supports controlled bursts while maintaining average rate limits. Most widely used. 5. Leaky Bucket Processes requests at a fixed outflow rate. Smooths bursts by queuing or dropping excess traffic. Predictable but less flexible. 6. Concurrency Limiter Limits how many requests run simultaneously - not per time window. Essential for protecting downstream services from overload. How to choose: → Need simplicity? Fixed Window → Need accuracy? Sliding Window Log → Need balance? Sliding Window Counter → Need burst tolerance? Token Bucket → Need smooth throughput? Leaky Bucket → Protecting a slow backend? Concurrency Limiter Most production systems combine 2–3 of these at different layers - gateway, service, and database. One algorithm rarely covers all your attack surfaces. Which one does your system rely on? 👇
-
Sidney Jones liked thisSidney Jones liked thisRecommendations for high-intermediate AI users: 1) Use Claude Code in the desktop app, not the command line app. It's fine to use the command line app sometimes, and it's used underneath the desktop app, anyway, but the desktop app provides a good 'session manager.' You can have sessions that are local, remote via ssh, or in a cloud env. The cloud envs are awesome, because there's way less risk of messing up something locally, which also means Claude can be more aggressive about doing things without asking. It's also easier when you jump between Antigravity/Cursor and Claude/ChatGPT, because you don't as often end up with local stuff in one terminal environment where your CL Claude Code is running that nothing else can see (like env vars, for example). 2) Use an external context manager (like miriel) with your Claude Code/Cowork or your ChatGPT/Codex, and with your Github Copilot. One of the great benefits is that it can be shared between all of them. Another is that it outlives sessions and YOU can manage the context (like insert random things from outside Claude Code, for example). My Claude Code kept making the same mistake. After having it fix the mistake for about the 5th time, I asked it to write a note I could give back to it in the future to avoid the same mistake. It wrote: "NEVER use variant="scrollable" on MUI Tabs components — it causes offsetHeight crashes during SSR." I then used my default Miriel Concept Manager 'tool' (MCP server) by saying: "add that note to my miriel project called 'dev gotchas.'" Now, any time I'm using any of these coding agents, they can use the same Miriel MCP server to 'recall' (the 'query' function in Miriel) that info, whether in a brand new conversation, or whatever. All things that support tool calling (even random Chinese models in LM Studio) can now see the 'dev gotchas' note. 3) the horizon for what the agents can do has lengthened greatly. Last night I had Claude Code add 27 Oauth integrations, change some output 'formatters,' fix a scrolling issue, re-do some dashboard overview cards, and refactor some code, all in one big prompt. I did the work of talking to it about the plan first, before having it start just before I went to sleep. It ended up being a 3-shot (two rounds of quick revisions required), but it worked well. This was not working 3 months ago. It is now. I now routinely say "add this feature, and make this page mobile friendly, and re-factor X and Y, and also fix this persistent issue by making the code more defensive, and factor out that UI element and replicate it on another page" whereas I used to have to do those things one at a time.
Experience & Education
-
University of Phoenix
View Sidney’s full experience
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Licenses & Certifications
Volunteer Experience
-
Mentor
Compuware Community Mentoring Program
- 2 years
Education
Courses
-
Critical Thinking
321
Languages
-
English
-
Organizations
-
Automation Alley
Member
- Present -
Lifetime Fitness
Member
- Present
Other similar profiles
Explore more posts
-
Saul Jaramillo
Anthropic just dropped Claude Code Review → multi-agent swarm auto-scans every PR for bugs, verifies findings, ranks severity. Internal stat: substantive reviews jumped 16% → 54%, catches ~7.5 real issues on big PRs. $15–25/review beta for Team/Ent. Game changer or expensive lint? 👀 https://lnkd.in/eVpR_vyD
-
Christian R. Vozar
Some thinking on Evals. I think it’s dangerously close to becoming so overloaded as a term that it obscures more than it clarifies. When doing what they do, evals should function like traditional software tests. Prevent regressions, speed up iteration, and provide sanity checks before shipping. The limitations are structural and by definition, they only reflect the failure cases we already know about. In environments where models are non-deterministic, frequently updated, and increasingly agentic, real truth appears to only emerge in production. Proper monitoring, experimentation, and error analysis in live systems can provide ground truth signals that evals cannot capture, especially when personalization fragments the definition of “good” into millions of micro-contexts. Evals have a role, but the future is not eval-driven development (please, God) but production-informed development, where small sets of smoke tests are paired with rich, real-time monitoring to surface what test suites can’t predict.
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content