Published:
Middle East Technical University’s sailing club has long been a hub for students who love the sea. I joined the club about six years ago, eager to escape Ankara’s dry inland climate, and spent nearly every free weekend on the Aegean. I’m not a member today, but the lessons still stick. Learning to call sail trim and feeling the hull surge after a clean tack gave me my first taste of real teamwork.
Published:
Middle East Technical University’s sailing club has long been a hub for students who love the sea. I joined the club about six years ago, eager to escape Ankara’s dry inland climate, and spent nearly every free weekend on the Aegean. I’m not a member today, but the lessons still stick. Learning to call sail trim and feeling the hull surge after a clean tack gave me my first taste of real teamwork.
Published:
Middle East Technical University’s sailing club has long been a hub for students who love the sea. I joined the club about six years ago, eager to escape Ankara’s dry inland climate, and spent nearly every free weekend on the Aegean. I’m not a member today, but the lessons still stick. Learning to call sail trim and feeling the hull surge after a clean tack gave me my first taste of real teamwork.
Published:
Entrepreneurs, researchers, and engineers live in a torrent of guidance. Podcasts, newsletters, and mentors offer conflicting prescriptions, each delivered with confidence. Distinguishing signal from noise is therefore a crucial skill. The discipline of evidence‑based management argues that decisions should be grounded in the best available data rather than authority or habit (Pfeffer & Sutton, 2006). Applying this mindset to advice means scrutinizing both the source and the context before acting.
Published:
Traditional agile boards depend on centralized tools and manual updates. As projects span multiple organizations and trust boundaries, these boards struggle to provide real-time, trustworthy visibility. Blockchain oracles and smart contracts offer a path to an autonomous, tamper-evident coordination layer that could serve as the world’s agile board.
Published:
Every engineer knows the temptation of the “five-minute hack.” A bug appears, the schedule is tight, and a clever shortcut promises to save the day. Weeks later, the team is still wrestling with the fallout. The myth of the quick fix lies in its invisibility—short-term patches often masquerade as efficient solutions while secretly accruing interest that compounds over time.
Published:
Consider this your two-minute hangar brief. Between February 2019 and November 2020, I served as an Expert Software Engineer at STM Defence Technologies in Ankara. My team and I wrote the operational brains for Kargu, a loitering munition that can patiently orbit before striking, and Togan, a reconnaissance UAV that refuses to land until it has found every point of interest. Coffee was strong, timelines were tighter than rotor tolerances, and every build had to keep aircrews safe.
Published:
When you strap into a Level D full flight simulator (FFS), you are basically sitting in a certified lie. Every cue—visual, auditory, motion, and the grumpy hydraulic groan when you bank too hard—is required to be indistinguishable from the real cockpit. During my stint as a Senior Software Engineer at Avion (2023–2025), I lived inside that lie for almost two years, shoulder to shoulder with test pilots, avionics engineers, and regulators. Our job? Make sure that when the sim sneezed, the aircraft would have sneezed too.
Published:
It started while I was nursing a Saturday-morning coffee and catching up on YouTube. Around the 11:20 mark of this talk, the speaker described the choreography of DNA replication—how strands of genetic code carry instructions that produce the very enzymes, such as DNA polymerase, needed to copy themselves. The diagram on screen sparked an immediate flashback to the time I first wrangled the Y combinator into a Haskell lab assignment. Suddenly, the elegance of Church’s abstraction and the biology lecture overlapped in my head: both were telling a story about systems that bootstrap their own continuation.
Published:
Machine learning (ML) is increasingly used to make blockchain networks more secure, efficient, and user-friendly. When I co-authored our survey on blockchain-enhanced machine learning [1], I kept hearing founders ask the inverse question: “How can ML shore up my ledger right now?” This post collects the playbooks that stuck with me from those conversations and the lab prototypes we built afterward.
Published:
Traditional agile boards depend on centralized tools and manual updates. As projects span multiple organizations and trust boundaries, these boards struggle to provide real-time, trustworthy visibility. Blockchain oracles and smart contracts offer a path to an autonomous, tamper-evident coordination layer that could serve as the world’s agile board.
Published:
Blockchain technology and machine learning (ML) have traditionally evolved on parallel tracks. However, recent research shows that their convergence promises transparent, trustworthy, and decentralized intelligent systems. In our survey on blockchain-enhanced machine learning [1], we examine how the decentralized design of blockchain can address systemic challenges in ML and pave the way for collaborative, verifiable artificial intelligence.
Published:
Living as an expat has given me the chance to compare daily life in the United States and the Netherlands in a way that guidebooks never could. My notebook is full of tiny contrasts: Leiden grocery receipts, NS train schedules for my commute, and lab to-do lists from Daytona Beach.
Published:
It started while I was nursing a Saturday-morning coffee and catching up on YouTube. Around the 11:20 mark of this talk, the speaker described the choreography of DNA replication—how strands of genetic code carry instructions that produce the very enzymes, such as DNA polymerase, needed to copy themselves. The diagram on screen sparked an immediate flashback to the time I first wrangled the Y combinator into a Haskell lab assignment. Suddenly, the elegance of Church’s abstraction and the biology lecture overlapped in my head: both were telling a story about systems that bootstrap their own continuation.
Published:
Machine learning (ML) is increasingly used to make blockchain networks more secure, efficient, and user-friendly. When I co-authored our survey on blockchain-enhanced machine learning [1], I kept hearing founders ask the inverse question: “How can ML shore up my ledger right now?” This post collects the playbooks that stuck with me from those conversations and the lab prototypes we built afterward.
Published:
Lean engineering teams face a paradox: they need specialists to build complex products, but they also need people who can switch contexts rapidly as priorities shift. Generalists thrive in this environment. They fill gaps no job description anticipated and create connective tissue between disciplines.
Published:
Living as an expat has given me the chance to compare daily life in the United States and the Netherlands in a way that guidebooks never could. My notebook is full of tiny contrasts: Leiden grocery receipts, NS train schedules for my commute, and lab to-do lists from Daytona Beach.
Published:
Working on simulator software, defensive cybersecurity platforms, and doctoral research has taught me that durable progress rarely starts with a perfect roadmap. It starts with persistent curiosity. Loewenstein’s review of the psychology of curiosity describes it as a response to information gaps that push humans to seek missing data rather than settle for superficial explanations (Loewenstein, 1994). On the Avion telemetry team those gaps appeared as cockpit workflows that felt clumsy or metrics that refused to line up with pilot reports. Pulling on those loose threads—asking “why does this feel wrong?”—often exposed real product opportunities.
Published:
Blockchain technology and machine learning (ML) have traditionally evolved on parallel tracks. However, recent research shows that their convergence promises transparent, trustworthy, and decentralized intelligent systems. In our survey on blockchain-enhanced machine learning [1], we examine how the decentralized design of blockchain can address systemic challenges in ML and pave the way for collaborative, verifiable artificial intelligence.
Published:
Entrepreneurs, researchers, and engineers live in a torrent of guidance. Podcasts, newsletters, and mentors offer conflicting prescriptions, each delivered with confidence. Distinguishing signal from noise is therefore a crucial skill. The discipline of evidence‑based management argues that decisions should be grounded in the best available data rather than authority or habit (Pfeffer & Sutton, 2006). Applying this mindset to advice means scrutinizing both the source and the context before acting.
Published:
Consider this your two-minute hangar brief. Between February 2019 and November 2020, I served as an Expert Software Engineer at STM Defence Technologies in Ankara. My team and I wrote the operational brains for Kargu, a loitering munition that can patiently orbit before striking, and Togan, a reconnaissance UAV that refuses to land until it has found every point of interest. Coffee was strong, timelines were tighter than rotor tolerances, and every build had to keep aircrews safe.
Published:
Before a developer ever sees your landing page, they might encounter your README. For many technical products, documentation is the first user interface. It guides users through setup, communicates architecture decisions, and sets expectations for how the system behaves. Neglecting documentation is like shipping a GUI with missing buttons and unlabeled fields.
Published:
Engineering teams love new tools. The promise of faster development, cleaner abstractions, or a more elegant stack is hard to resist. Yet every adoption decision carries hidden costs that rarely appear in procurement spreadsheets. The real price of a shiny tool lies in the cognitive load it introduces, the training it demands, and the fragile glue it creates between systems.
Published:
Before a developer ever sees your landing page, they might encounter your README. For many technical products, documentation is the first user interface. It guides users through setup, communicates architecture decisions, and sets expectations for how the system behaves. Neglecting documentation is like shipping a GUI with missing buttons and unlabeled fields.
Published:
Consider this your two-minute hangar brief. Between February 2019 and November 2020, I served as an Expert Software Engineer at STM Defence Technologies in Ankara. My team and I wrote the operational brains for Kargu, a loitering munition that can patiently orbit before striking, and Togan, a reconnaissance UAV that refuses to land until it has found every point of interest. Coffee was strong, timelines were tighter than rotor tolerances, and every build had to keep aircrews safe.
Published:
Every engineer knows the temptation of the “five-minute hack.” A bug appears, the schedule is tight, and a clever shortcut promises to save the day. Weeks later, the team is still wrestling with the fallout. The myth of the quick fix lies in its invisibility—short-term patches often masquerade as efficient solutions while secretly accruing interest that compounds over time.
Published:
Embarking on a PhD in Florida means waking up to salt air, sunscreen reminders, and the low rumble of aircraft testing engines at Embry-Riddle Aeronautical University. My advisor jokes that our campus smells like jet fuel and ambition; he’s right on both counts. By 7 a.m. I am usually rolling past the flight line with a notebook full of model checkpoints to validate and a thermos of Cuban coffee strong enough to double as rocket propellant.
Published:
Living as an expat has given me the chance to compare daily life in the United States and the Netherlands in a way that guidebooks never could. My notebook is full of tiny contrasts: Leiden grocery receipts, NS train schedules for my commute, and lab to-do lists from Daytona Beach.
Published:
Working on simulator software, defensive cybersecurity platforms, and doctoral research has taught me that durable progress rarely starts with a perfect roadmap. It starts with persistent curiosity. Loewenstein’s review of the psychology of curiosity describes it as a response to information gaps that push humans to seek missing data rather than settle for superficial explanations (Loewenstein, 1994). On the Avion telemetry team those gaps appeared as cockpit workflows that felt clumsy or metrics that refused to line up with pilot reports. Pulling on those loose threads—asking “why does this feel wrong?”—often exposed real product opportunities.
Published:
Embarking on a PhD in Florida means waking up to salt air, sunscreen reminders, and the low rumble of aircraft testing engines at Embry-Riddle Aeronautical University. My advisor jokes that our campus smells like jet fuel and ambition; he’s right on both counts. By 7 a.m. I am usually rolling past the flight line with a notebook full of model checkpoints to validate and a thermos of Cuban coffee strong enough to double as rocket propellant.
Published:
Lean engineering teams face a paradox: they need specialists to build complex products, but they also need people who can switch contexts rapidly as priorities shift. Generalists thrive in this environment. They fill gaps no job description anticipated and create connective tissue between disciplines.
Published:
It started while I was nursing a Saturday-morning coffee and catching up on YouTube. Around the 11:20 mark of this talk, the speaker described the choreography of DNA replication—how strands of genetic code carry instructions that produce the very enzymes, such as DNA polymerase, needed to copy themselves. The diagram on screen sparked an immediate flashback to the time I first wrangled the Y combinator into a Haskell lab assignment. Suddenly, the elegance of Church’s abstraction and the biology lecture overlapped in my head: both were telling a story about systems that bootstrap their own continuation.
Published:
Proof of Learning (PoL) verifies that a model was genuinely trained on claimed data by providing verifiable evidence of the training process. I first felt the urgency for PoL while helping an aerospace partner document how a safety-critical model was trained—our counterparts were less interested in raw accuracy and more concerned about whether they could audit each gradient step. As machine learning systems become pervasive in critical domains, PoL offers a mechanism to ensure trust and accountability in model provenance.
Published:
Embarking on a PhD in Florida means waking up to salt air, sunscreen reminders, and the low rumble of aircraft testing engines at Embry-Riddle Aeronautical University. My advisor jokes that our campus smells like jet fuel and ambition; he’s right on both counts. By 7 a.m. I am usually rolling past the flight line with a notebook full of model checkpoints to validate and a thermos of Cuban coffee strong enough to double as rocket propellant.
Published:
Model watermarking embeds identifiable patterns into a model’s parameters or outputs so that ownership can be demonstrated without access to the original training process. Balancing my research life with day-to-day engineering at Avion Full Flight Simulators means I constantly straddle theory and high-stakes delivery. I still remember the first time we recovered our watermark from a partner’s heavily fine-tuned flight-control model inside the Avion Cloud dashboard—it felt like catching a subtle watermark in a printed banknote after a white-knuckle certification sprint. That breakthrough was grounded in early work showing that deep network weights can carry hidden signatures without affecting accuracy [4], and in behavior-based marks that survive aggressive fine-tuning [5,6].
Published:
Machine learning (ML) is increasingly used to make blockchain networks more secure, efficient, and user-friendly. When I co-authored our survey on blockchain-enhanced machine learning [1], I kept hearing founders ask the inverse question: “How can ML shore up my ledger right now?” This post collects the playbooks that stuck with me from those conversations and the lab prototypes we built afterward.
Published:
Blockchain technology and machine learning (ML) have traditionally evolved on parallel tracks. However, recent research shows that their convergence promises transparent, trustworthy, and decentralized intelligent systems. In our survey on blockchain-enhanced machine learning [1], we examine how the decentralized design of blockchain can address systemic challenges in ML and pave the way for collaborative, verifiable artificial intelligence.
Published:
Earlier this year I moved to a new neighborhood and was stunned by how long those first weeks felt. Every grocery run, bus ride, and “where on earth do they hide the trash bags?” errand demanded full attention. The days stretched like taffy. Six weeks later, the same block felt routine and the months started blurring together again. Apparently it is not just the candles on my birthday cake—cognitive science says the culprit is the way my brain compresses sameness and bookmarks surprises12.
Avni-Babad, D., & Ritov, I. (2003). Routine and the perception of time. Journal of Experimental Psychology: General, 132(4), 543–553. https://doi.org/10.1037/0096-3445.132.4.543 ↩
Kurby, C.A., & Zacks, J.M. (2008). Segmentation in the perception and memory of events. Trends in Cognitive Sciences, 12(2), 72–79. https://doi.org/10.1016/j.tics.2007.11.004 ↩
Published:
Model watermarking embeds identifiable patterns into a model’s parameters or outputs so that ownership can be demonstrated without access to the original training process. Balancing my research life with day-to-day engineering at Avion Full Flight Simulators means I constantly straddle theory and high-stakes delivery. I still remember the first time we recovered our watermark from a partner’s heavily fine-tuned flight-control model inside the Avion Cloud dashboard—it felt like catching a subtle watermark in a printed banknote after a white-knuckle certification sprint. That breakthrough was grounded in early work showing that deep network weights can carry hidden signatures without affecting accuracy [4], and in behavior-based marks that survive aggressive fine-tuning [5,6].
Published:
Earlier this year I moved to a new neighborhood and was stunned by how long those first weeks felt. Every grocery run, bus ride, and “where on earth do they hide the trash bags?” errand demanded full attention. The days stretched like taffy. Six weeks later, the same block felt routine and the months started blurring together again. Apparently it is not just the candles on my birthday cake—cognitive science says the culprit is the way my brain compresses sameness and bookmarks surprises12.
Avni-Babad, D., & Ritov, I. (2003). Routine and the perception of time. Journal of Experimental Psychology: General, 132(4), 543–553. https://doi.org/10.1037/0096-3445.132.4.543 ↩
Kurby, C.A., & Zacks, J.M. (2008). Segmentation in the perception and memory of events. Trends in Cognitive Sciences, 12(2), 72–79. https://doi.org/10.1016/j.tics.2007.11.004 ↩
Published:
Before a developer ever sees your landing page, they might encounter your README. For many technical products, documentation is the first user interface. It guides users through setup, communicates architecture decisions, and sets expectations for how the system behaves. Neglecting documentation is like shipping a GUI with missing buttons and unlabeled fields.
Published:
Traditional agile boards depend on centralized tools and manual updates. As projects span multiple organizations and trust boundaries, these boards struggle to provide real-time, trustworthy visibility. Blockchain oracles and smart contracts offer a path to an autonomous, tamper-evident coordination layer that could serve as the world’s agile board.
Published:
Embarking on a PhD in Florida means waking up to salt air, sunscreen reminders, and the low rumble of aircraft testing engines at Embry-Riddle Aeronautical University. My advisor jokes that our campus smells like jet fuel and ambition; he’s right on both counts. By 7 a.m. I am usually rolling past the flight line with a notebook full of model checkpoints to validate and a thermos of Cuban coffee strong enough to double as rocket propellant.
Published:
Scaling conversations dominate startup culture, yet the first question any product must answer is painfully small: will even one person use it when given the chance? The only reliable way to find out is through direct observation. This principle is at the heart of evidence‑based design, a methodology grounded in human‑computer interaction research that emphasizes empirical feedback over speculation (Nielsen, 1993). Surveys and interviews hint at preferences, but behavior reveals intent. If your prototype cannot hold the attention of a single user, no amount of marketing will redeem it.
Published:
Product launches often steal the spotlight, but the feedback gathered during early beta testing quietly determines whether launch day is a victory lap or a scramble to patch obvious gaps. Treating beta periods as a checkbox diminishes their power; when leveraged well, beta feedback drives the evolution of features, design, and infrastructure before they congeal into something hard to change.
Published:
Engineering teams love new tools. The promise of faster development, cleaner abstractions, or a more elegant stack is hard to resist. Yet every adoption decision carries hidden costs that rarely appear in procurement spreadsheets. The real price of a shiny tool lies in the cognitive load it introduces, the training it demands, and the fragile glue it creates between systems.
Published:
Proof of Learning (PoL) verifies that a model was genuinely trained on claimed data by providing verifiable evidence of the training process. I first felt the urgency for PoL while helping an aerospace partner document how a safety-critical model was trained—our counterparts were less interested in raw accuracy and more concerned about whether they could audit each gradient step. As machine learning systems become pervasive in critical domains, PoL offers a mechanism to ensure trust and accountability in model provenance.
Published:
Middle East Technical University’s sailing club has long been a hub for students who love the sea. I joined the club about six years ago, eager to escape Ankara’s dry inland climate, and spent nearly every free weekend on the Aegean. I’m not a member today, but the lessons still stick. Learning to call sail trim and feeling the hull surge after a clean tack gave me my first taste of real teamwork.
Published:
Middle East Technical University’s sailing club has long been a hub for students who love the sea. I joined the club about six years ago, eager to escape Ankara’s dry inland climate, and spent nearly every free weekend on the Aegean. I’m not a member today, but the lessons still stick. Learning to call sail trim and feeling the hull surge after a clean tack gave me my first taste of real teamwork.
Published:
Proof of Learning (PoL) verifies that a model was genuinely trained on claimed data by providing verifiable evidence of the training process. I first felt the urgency for PoL while helping an aerospace partner document how a safety-critical model was trained—our counterparts were less interested in raw accuracy and more concerned about whether they could audit each gradient step. As machine learning systems become pervasive in critical domains, PoL offers a mechanism to ensure trust and accountability in model provenance.
Published:
Embarking on a PhD in Florida means waking up to salt air, sunscreen reminders, and the low rumble of aircraft testing engines at Embry-Riddle Aeronautical University. My advisor jokes that our campus smells like jet fuel and ambition; he’s right on both counts. By 7 a.m. I am usually rolling past the flight line with a notebook full of model checkpoints to validate and a thermos of Cuban coffee strong enough to double as rocket propellant.
Published:
Model watermarking embeds identifiable patterns into a model’s parameters or outputs so that ownership can be demonstrated without access to the original training process. Balancing my research life with day-to-day engineering at Avion Full Flight Simulators means I constantly straddle theory and high-stakes delivery. I still remember the first time we recovered our watermark from a partner’s heavily fine-tuned flight-control model inside the Avion Cloud dashboard—it felt like catching a subtle watermark in a printed banknote after a white-knuckle certification sprint. That breakthrough was grounded in early work showing that deep network weights can carry hidden signatures without affecting accuracy [4], and in behavior-based marks that survive aggressive fine-tuning [5,6].
Published:
Machine learning (ML) is increasingly used to make blockchain networks more secure, efficient, and user-friendly. When I co-authored our survey on blockchain-enhanced machine learning [1], I kept hearing founders ask the inverse question: “How can ML shore up my ledger right now?” This post collects the playbooks that stuck with me from those conversations and the lab prototypes we built afterward.
Published:
Blockchain technology and machine learning (ML) have traditionally evolved on parallel tracks. However, recent research shows that their convergence promises transparent, trustworthy, and decentralized intelligent systems. In our survey on blockchain-enhanced machine learning [1], we examine how the decentralized design of blockchain can address systemic challenges in ML and pave the way for collaborative, verifiable artificial intelligence.
Published:
When you strap into a Level D full flight simulator (FFS), you are basically sitting in a certified lie. Every cue—visual, auditory, motion, and the grumpy hydraulic groan when you bank too hard—is required to be indistinguishable from the real cockpit. During my stint as a Senior Software Engineer at Avion (2023–2025), I lived inside that lie for almost two years, shoulder to shoulder with test pilots, avionics engineers, and regulators. Our job? Make sure that when the sim sneezed, the aircraft would have sneezed too.
Published:
Traditional agile boards depend on centralized tools and manual updates. As projects span multiple organizations and trust boundaries, these boards struggle to provide real-time, trustworthy visibility. Blockchain oracles and smart contracts offer a path to an autonomous, tamper-evident coordination layer that could serve as the world’s agile board.
Published:
Entrepreneurs, researchers, and engineers live in a torrent of guidance. Podcasts, newsletters, and mentors offer conflicting prescriptions, each delivered with confidence. Distinguishing signal from noise is therefore a crucial skill. The discipline of evidence‑based management argues that decisions should be grounded in the best available data rather than authority or habit (Pfeffer & Sutton, 2006). Applying this mindset to advice means scrutinizing both the source and the context before acting.
Published:
Scaling conversations dominate startup culture, yet the first question any product must answer is painfully small: will even one person use it when given the chance? The only reliable way to find out is through direct observation. This principle is at the heart of evidence‑based design, a methodology grounded in human‑computer interaction research that emphasizes empirical feedback over speculation (Nielsen, 1993). Surveys and interviews hint at preferences, but behavior reveals intent. If your prototype cannot hold the attention of a single user, no amount of marketing will redeem it.
Published:
Working on simulator software, defensive cybersecurity platforms, and doctoral research has taught me that durable progress rarely starts with a perfect roadmap. It starts with persistent curiosity. Loewenstein’s review of the psychology of curiosity describes it as a response to information gaps that push humans to seek missing data rather than settle for superficial explanations (Loewenstein, 1994). On the Avion telemetry team those gaps appeared as cockpit workflows that felt clumsy or metrics that refused to line up with pilot reports. Pulling on those loose threads—asking “why does this feel wrong?”—often exposed real product opportunities.
Published:
Lean engineering teams face a paradox: they need specialists to build complex products, but they also need people who can switch contexts rapidly as priorities shift. Generalists thrive in this environment. They fill gaps no job description anticipated and create connective tissue between disciplines.
Published:
Machine learning (ML) is increasingly used to make blockchain networks more secure, efficient, and user-friendly. When I co-authored our survey on blockchain-enhanced machine learning [1], I kept hearing founders ask the inverse question: “How can ML shore up my ledger right now?” This post collects the playbooks that stuck with me from those conversations and the lab prototypes we built afterward.
Published:
Blockchain technology and machine learning (ML) have traditionally evolved on parallel tracks. However, recent research shows that their convergence promises transparent, trustworthy, and decentralized intelligent systems. In our survey on blockchain-enhanced machine learning [1], we examine how the decentralized design of blockchain can address systemic challenges in ML and pave the way for collaborative, verifiable artificial intelligence.
Published:
Every engineer knows the temptation of the “five-minute hack.” A bug appears, the schedule is tight, and a clever shortcut promises to save the day. Weeks later, the team is still wrestling with the fallout. The myth of the quick fix lies in its invisibility—short-term patches often masquerade as efficient solutions while secretly accruing interest that compounds over time.
Published:
Scaling conversations dominate startup culture, yet the first question any product must answer is painfully small: will even one person use it when given the chance? The only reliable way to find out is through direct observation. This principle is at the heart of evidence‑based design, a methodology grounded in human‑computer interaction research that emphasizes empirical feedback over speculation (Nielsen, 1993). Surveys and interviews hint at preferences, but behavior reveals intent. If your prototype cannot hold the attention of a single user, no amount of marketing will redeem it.
Published:
Product launches often steal the spotlight, but the feedback gathered during early beta testing quietly determines whether launch day is a victory lap or a scramble to patch obvious gaps. Treating beta periods as a checkbox diminishes their power; when leveraged well, beta feedback drives the evolution of features, design, and infrastructure before they congeal into something hard to change.
Published:
Earlier this year I moved to a new neighborhood and was stunned by how long those first weeks felt. Every grocery run, bus ride, and “where on earth do they hide the trash bags?” errand demanded full attention. The days stretched like taffy. Six weeks later, the same block felt routine and the months started blurring together again. Apparently it is not just the candles on my birthday cake—cognitive science says the culprit is the way my brain compresses sameness and bookmarks surprises12.
Avni-Babad, D., & Ritov, I. (2003). Routine and the perception of time. Journal of Experimental Psychology: General, 132(4), 543–553. https://doi.org/10.1037/0096-3445.132.4.543 ↩
Kurby, C.A., & Zacks, J.M. (2008). Segmentation in the perception and memory of events. Trends in Cognitive Sciences, 12(2), 72–79. https://doi.org/10.1016/j.tics.2007.11.004 ↩
Published:
Engineering teams love new tools. The promise of faster development, cleaner abstractions, or a more elegant stack is hard to resist. Yet every adoption decision carries hidden costs that rarely appear in procurement spreadsheets. The real price of a shiny tool lies in the cognitive load it introduces, the training it demands, and the fragile glue it creates between systems.
Published:
When you strap into a Level D full flight simulator (FFS), you are basically sitting in a certified lie. Every cue—visual, auditory, motion, and the grumpy hydraulic groan when you bank too hard—is required to be indistinguishable from the real cockpit. During my stint as a Senior Software Engineer at Avion (2023–2025), I lived inside that lie for almost two years, shoulder to shoulder with test pilots, avionics engineers, and regulators. Our job? Make sure that when the sim sneezed, the aircraft would have sneezed too.
Published:
Product launches often steal the spotlight, but the feedback gathered during early beta testing quietly determines whether launch day is a victory lap or a scramble to patch obvious gaps. Treating beta periods as a checkbox diminishes their power; when leveraged well, beta feedback drives the evolution of features, design, and infrastructure before they congeal into something hard to change.
Published:
It started while I was nursing a Saturday-morning coffee and catching up on YouTube. Around the 11:20 mark of this talk, the speaker described the choreography of DNA replication—how strands of genetic code carry instructions that produce the very enzymes, such as DNA polymerase, needed to copy themselves. The diagram on screen sparked an immediate flashback to the time I first wrangled the Y combinator into a Haskell lab assignment. Suddenly, the elegance of Church’s abstraction and the biology lecture overlapped in my head: both were telling a story about systems that bootstrap their own continuation.