OldKnowledge,New Vessel.
Essays on on-device AI, data sovereignty, and building systems that keep knowledge where it belongs.
Enter the archive →— 00 / ThesisEvery business runs on knowledge older than its software.
Latest Entries
Designing for the DMA: Interoperability Is an API Product With a Regulator in the Room, So Build the Access Layer Once (2026)
Continual Learning in Production: A Model That Keeps Learning Is a Release Train, So Gate Every Update Like a Deploy (2026)
Sim-to-Real Transfer Architecture: The Reality Gap Is a Measurement, Not a Feeling, So Budget It Before You Randomise It (2026)
Continuous Profiling in Production: The Profile Is the Only Telemetry Denominated in Money, So Run It as a Cost Ledger First and a Debugger Second (2026)
Terraform at Platform Scale: State Is the Unit of Blast Radius, So Split It by Who Changes What and How Often, Not by Environment (2026)
अथ प्रथमोऽध्यायः ॥
विद्या ददाति विनयं विनयाद् याति पात्रताम् ।
पात्रत्वाद्धनमाप्नोति धनाद्धर्मं ततः सुखम् ॥
We digitize centuries-old manuscripts. Then we build with the same discipline.
AppScale's roots are in a quiet project: structuring classical Sanskrit texts into faithful digital form. Extraction, structure, provenance, sovereignty — the same principles now power our client work.
Built for Every Business
Intelligence that never leaves the device
Offline-first, data-sovereign systems — LLMs, vector search, inference at the edge.
02Document IntelligenceAir-gapped document vaults
Sealed, searchable, provable.
03AI ProductsFull-cycle AI product builds
Mobile-first, production-grade.
04MCP & AI SecuritySecuring the agentic stack
Tool poisoning, prompt injection, confused deputies — studied and hardened.
05Indic Language AIOCR for scripts the world forgot
Sovereign OCR for Indic scripts.
06Consulting & Architecture19+ years shipping at scale
OTT, e-commerce & mobility platforms serving millions.
Designing for the DMA: Interoperability Is an API Product With a Regulator in the Room, So Build the Access Layer Once (2026)
DMA interoperability in 2026: one access layer for identity, consent, scoped tokens, integrity rules and audit, plus how third parties design for slow access.
Continual Learning in Production: A Model That Keeps Learning Is a Release Train, So Gate Every Update Like a Deploy (2026)
Continual learning in 2026: every model update is a release. Gate it on a frozen memory set, use replay against catastrophic forgetting, keep a rollback.
Sim-to-Real Transfer Architecture: The Reality Gap Is a Measurement, Not a Feeling, So Budget It Before You Randomise It (2026)
Sim-to-real transfer in 2026: the reality gap is a measurement, not a feeling. Decompose it, calibrate what you can, randomise the rest, gate on hardware.
Continuous Profiling in Production: The Profile Is the Only Telemetry Denominated in Money, So Run It as a Cost Ledger First and a Debugger Second (2026)
Continuous profiling in 2026: profiles are the only telemetry denominated in money. Run eBPF sampling as a fleet cost ledger first, a debugger second.
Terraform at Platform Scale: State Is the Unit of Blast Radius, So Split It by Who Changes What and How Often, Not by Environment (2026)
How to structure Terraform for many teams in 2026: state is the unit of blast radius, so layer it by owner and change cadence, then pick Stacks or Terragrunt.
Islands Architecture and Partial Hydration in 2026: Hydration Is a Per-Component Tax, So Your Interactivity Ratio Picks the Architecture, Not the Framework
Islands vs React Server Components vs Qwik in 2026: hydration is a per-component tax, so measure your interactivity ratio before you pick an architecture.
Secure Multi-Party Computation for Cross-Organisation Analytics: The Protocol Is the Easy Part, the Non-Collusion Contract Is the Product (2026)
Secure multi-party computation in production in 2026: why non-collusion governance matters more than the protocol, how to architect it and when to pick a cheaper tool.
Safety Certification for AI-Driven Robots: Certify the Envelope, Not the Policy, and Keep Learning Out of the Safety Function (2026)
How to certify an AI-driven robot under the EU Machinery Regulation in 2026: keep machine learning out of the safety function and certify a deterministic envelope.
Platform Team Funding and Staffing: Chargeback Taxes Adoption, So Fund the Floor, Meter the Variable and Never Bill the Paved Road (2026)
How to fund and staff a platform team in 2026: why chargeback taxes adoption, the fund-the-floor hybrid model, bottom-up headcount and what to report to finance.
Robot Teleoperation Fallback Architecture: The Remote Human Should Advise, Not Drive, and the Operator Ratio Is Your Real Business Model (2026)
Robot teleoperation fallback done right: the robot owns safety, remote operators advise rather than drive, and the operator ratio is your real business model.
Differential Privacy in Production, 2026: Epsilon Is a Budget, Not a Setting, and Deployments Fail at the Ledger, Not the Math (2026)
Differential privacy in production fails at the ledger, not the math. How to budget epsilon per dataset, gate queries, use NIST SP 800-226, and when to skip DP.
AI Weather Models in Production, 2026: The Speedup Is an Ensemble Dividend, and You Only Collect It If Your Decisions Consume Probabilities (2026)
AI weather models like ECMWF AIFS and WeatherNext 3 made ensembles cheap. Why the value is in calibrated probabilities, the pipeline, and when not to switch.
The View Transitions API in 2026: Cross-Document Transitions Refund the SPA Tax You Paid for Smooth Navigation, in Two Engines Out of Three (2026)
Cross-document view transitions give multi-page sites SPA-style navigation in Chrome and Safari. When to drop the SPA, the limits, and what Firefox users get.
vLLM vs SGLang vs TensorRT-LLM: Your Prefix-Reuse Ratio Picks the Inference Server, Not the Leaderboard (2026)
vLLM, SGLang or TensorRT-LLM? How to choose an LLM inference server in 2026 by prefix-reuse ratio and hardware commitment, not by a public benchmark.
Multi-Tenancy in Internal Platforms: Namespaces Isolate What Is Namespaced, So Choose the Tenancy Model by Who Owns the Cluster-Scoped Objects (2026)
Namespace per team, vCluster virtual control planes or cluster per team: how to pick a Kubernetes tenancy model for internal platforms by who owns the cluster-scoped objects.
Homomorphic Encryption in Production: The Server Computes Blind, So Only Fixed-Shape Lookups and Small Models Pay, and Everything Else Is a Trusted-Execution Decision (2026)
What homomorphic encryption can do in production in 2026, what Apple and Zama actually run, where it fails, and when a trusted execution environment is the honest answer.
Real-Time Streaming Into a Digital Twin: The Twin Is a Materialised View With a Staleness Contract, Not a Database You Write Sensor Data Into (2026)
Sparkplug, MQTT, Kafka and the twin platform in four tiers: how sensor data should flow into Azure Digital Twins, TwinMaker or Ditto, and the staleness contract that stops lag.
WASM at the Edge: The Cold Start Is the Price of the Trust Boundary, So Choose the Sandbox by What the Code Must Be Allowed to Do (2026)
Isolates, Wasm sandboxes, microVMs and containers compared on the limits vendors publish, where Wasm is the right edge boundary, and what no cold start really costs you.
World Models in Production: The Model Simulates the Environment, So Treat It as a Generative Test Harness and Never as Ground Truth (2026)
Genie 3, Cosmos 3, Waymo's world model and Marble: where a generative environment belongs in a robotics or autonomy stack, the two-simulator pattern, and where it lies to you.
Confidential Containers on Kubernetes: The Pod Treats Its Own Cluster as the Adversary, So Every Platform Convenience Becomes a Policy Decision (2026)
Confidential containers put the pod in a TEE and your own Kubelet, admins and control plane outside it. CoCo v0.23, AKS preview, GKE nodes and OpenShift 1.13 compared for 2026.
Bring this thinkingto your business.
conversation
One essay, most weeks. No noise.