Loading…
Latest reads
Every post, summarised
Grounded summaries of engineering blogs, newest first. Each links to the original.
- 4,671 posts
- 20 companies
The Good Stuff
Paid picks we use ourselves. Links carry a referral code — a subscription through one earns us a commission at no extra cost to you.
Latest posts
How Swiggy Saved 3.7 MB by Solving a Vector Drawable Mystery
An Android APK investigation found high-resolution PNG variants beside vector drawable XML files, although those raster assets had not been added directly to the codebase. The cause was AAPT2 rasterizing vectors that used features unavailable below API 24, including gradients in android:fillColor, despite the app’s minSdkVersion being 23; the generated files could be much larger than their XML sources, and were produced even for unused resources. The described remedy was to remove unused raster-prone assets, use VectorDrawableCompat with app:srcCompat for affected vectors, and tell Gradle and AAPT to use support-library rendering rather than density-specific PNGs. Removing unused assets cut download size by more than 1 MB, while raising minSdkVersion to 24 produced a further 2.7 MB reduction, for 3.7 MB total; similar gains were observed in Toing and Instamart.
How Swiggy Migrated from protobuf-java to Square Wire
Swiggy migrated from protobuf-java to Square Wire after profiling showed that the runtime and protoc-generated code consumed 6.735 MB of Android app size. protobuf-javalite was rejected because Swiggy relies extensively on google.protobuf.Any, while Wire retained support for Any and other well-known types with a smaller runtime and immutable Kotlin models. The migration required updating snake_case property access, message construction, serialization calls, ProGuard rules, and a custom pipeline that compiled all schemas together before publishing package-specific AARs to AWS CodeArtifact. The team also fixed repeated enum fields with [packed = true] after discovering a silent wire-format mismatch. It reduced Swiggy’s app size by 3.52 MB and produced measurable reductions for Instamart and Toing.
Scaling Android CI: From 44 → 10 minutes — A Deep Dive into Build Time Optimization and Best…
An Android monorepo’s CI pipeline averaged about 35 minutes in its build phase, with a P90 of roughly 44 minutes, as repeated dependency downloads, environment-driven invalidations, variant bloat, and machine limits increased work. The team reworked caching with branch- and architecture-aware keys, release/develop fallbacks, cached SDK and dependency data, and a Gradle Remote Cache, while filtering repositories and restricting CI to required variants. It also tuned 32GB runners to a 16GB heap and 12GB Metaspace, disabled configuration-cache saves in ephemeral CI, upgraded Kotlin, AGP 8.10.1, and Gradle 8.11.1, and moved critical workflows to Apple Silicon. Gradle tasks fell from 3399 to 2745, while remote-cache hit rates reached P50 44% and P90 56%. After the Apple Silicon refresh, total CI time reached P50 10 and P90 11 minutes, average credits per build dropped from 55 to 43, and a comparable rollout window saved about 5800 credits.
UST is bringing Claude to physical AI
UST is partnering with Anthropic to bring Claude into engineering and production systems used by semiconductor, automotive, manufacturing, telecom, embedded, and IoT companies. Claude Code reads chip pinouts and hardware schematics, then writes and runs regression tests while carrying design context across hours-long tasks. UST is integrating Claude as the reasoning layer in its iDEC validation pipeline, which generates tests, compares live equipment data with a digital twin, and already cuts validation cycles from four days to 48 hours, 50 to 70 percent. The partnership also applies Claude to healthcare, telecom, and banking platforms, with human approval retained for recommended actions and response workflows. UST will train 20,000 associates worldwide and build specialized deployment teams, with governance and audit controls supporting production use in high-stakes industries.
Side Chats and Conversation Search
An update adds side chats for exploring tangents alongside a main agent conversation without interrupting it. Each side chat is a durable, full agent conversation with main-chat context, can be revisited, followed up on, and at-mentioned back into the main thread. In the Agents Window, users can search agent transcripts through Cmd+K using a local index designed to scale to thousands of conversations, or search within a conversation with Cmd+F. Redesigned project and repo pickers support creating projects and connecting GitHub, GitLab, or Azure DevOps without leaving the picker, with search scoped to the current location. New cloud agent hooks expose prompts, responses, thinking, subagents, compaction, and turn completion for observation and control.
Scaling Grab's Data Lake: Our journey to Apache Iceberg adoption
Grab is transitioning its petabyte-scale data lake from Hive Parquet tables managed through the Hive Metastore to a table-centric architecture centered on Apache Iceberg, while retaining storage-compute separation on S3. The move addresses catalog listing latency, severe small-file fragmentation, manual partition registration, absent native ACID operations, and catalog state becoming stale after direct S3 changes. Rather than converting the entire lake at once, Grab prioritizes high-value tables and uses features such as Z-ordering; representative workloads cut one query from 70 seconds to 6, reduced daily S3 API costs by up to 95%, and halved cluster resource usage. Its open-sourced UnifiedSparkCatalog detects table formats through metastore metadata or path inference and routes operations to the appropriate Spark catalog without format-specific prefixes. Deployment also required resolving Hive lock contention, timestamp incompatibilities, and storage-tier cost spikes.
Ask, build, compose: What our 5th Genie Hackathon taught us about Databricks Genie
The fifth Databricks hackathon used Databricks Genie to test three ways of working with governed data: asking, building, and composing. Genie Agents give business users domain-specific natural-language access to curated data, while Genie Code helps analysts create metric views, functions, pipelines, and dashboards inside Databricks. The composition track uses Conversation APIs and a managed MCP server to make Genie an agent tool alongside custom logic, Model Serving, and other MCP servers, with Unity Catalog governing access. Projects included supervisory routing across 190 tables, governance and analytics products built in days, and multi-agent systems such as ShipBob’s overnight operations brief. Across the tracks, the stated lesson is that shared governance and semantic context let business users, builders, and engineers use Genie at different levels without abandoning grounded, permissioned data access.
From Day 1 to Production: Building Lyft’s Analytics & Rides Intelligence Assistant as Onboarding…
Lyft developed Aria, an AI-powered assistant allowing approved employees to query ride analytics via natural language through a backend with LangGraph orchestration and SQL generation. However, Aria's user interface was limited to a Streamlit prototype that lacked standard service framework integration, Lyft authentication, and multi-user scalability. To resolve these limitations, a new engineer built a production-grade web client from scratch using Lyft's internal Node.js framework and Next.js as a three-week onboarding assignment. The implementation involved configuring Envoy routing, resolving authentication plugin incompatibilities, debugging service connectivity with Grafana, and managing interface state with an XState state machine. Aria was successfully hardened and launched into production, expanding natural language analytics access across the company.
Scaling Verify with Wallet for Identity Verification at Uber
Uber integrated Apple’s Verify with Wallet API into its cross-app Identity Verification Platform to support government-backed digital IDs across driver onboarding, age-gated ordering, and car rental use cases. The central challenge was allowing each use case to request only its Apple-approved identity data while operating through a scalable, platform-agnostic service rather than separate integrations. Server-side configurations map Uber identity models to PassKit objects, set one-time or persistent storage intent, and validate responses using HPKE decryption, CBOR payloads, session transcripts, MSOs, device signatures, and issuer certificate chains under ISO/IEC 18013-5. A registry of IACA certificates, sourced from jurisdictional portals and AAMVA’s VICAL, supports overlapping certificate rotations, while the platform is intended to accommodate new document types and cross-platform digital credential integrations.
Golden Gate Claude
Golden Gate Claude presents a research demonstration of interpretability work on Claude 3 Sonnet. Researchers identified millions of activated concepts, or “features,” including a specific combination of neurons associated with the Golden Gate Bridge, and adjusted that feature’s activation strength. Increasing it caused responses to mention the bridge across unrelated prompts, such as spending $10 or writing a love story; the demonstration was available online for 24 hours and is no longer available. The post distinguishes this intervention from verbal role-play, system prompts, and traditional fine-tuning, describing it as a precise change to internal activations. It also says similar techniques could modify safety-related features connected to dangerous code, criminal activity, or deception, potentially supporting safer models.
Synthetic Data Generation for Financial AI Research with NVIDIA NeMo
Real-world financial natural language processing datasets often overrepresent common events like earnings while underrepresenting rarer occurrences such as credit-rating changes. To address this imbalance, an iterative generation pipeline produced over 500,000 unique headlines across 13 categories using NVIDIA NeMo Data Designer, NeMo Curator, and Nemotron 3 Nano. Instead of relying on a single large batch that yielded 65 percent near-duplicates, the workflow executes iterative cycles of category-weighted generation, rule-based filtering, global semantic deduplication, and dynamic category reweighting. Farthest-from-centroid few-shot example selection guided subsequent rounds toward novel semantic spaces across 82 total iterations. Executed over approximately six days on an eight-way NVIDIA B200 node, the resulting corpus enabled fine-tuning compact student language models to achieve 95 percent of teacher model performance.
The Long-Term Benefit Trust
Anthropic announces the Long-Term Benefit Trust (LTBT), a new governance structure intended to address the unusually large public externalities it associates with transformative AI. The independent Trust has five financially disinterested Trustees and, alongside Anthropic’s Delaware Public Benefit Corporation status, is meant to add accountability to the public interest without directing ordinary commercial decisions. Created through Class T stock after Series C, it can elect and remove board members, with authority phasing in through time- and funding-based milestones and reaching a board majority within four years; investors also receive a new directly elected seat. Protective provisions require notice of certain major actions, while amendment rules combine agreement-based changes with supermajority failsafes whose thresholds rise as the Trust gains power. Anthropic presents the arrangement as an experiment, with initial one-year Trustees selected after a year-long search and future Trustees elected by the Trustees.
Inviting hard questions
Anthropic is inviting the public to submit difficult questions about AI’s effects on jobs, society, families, science, medicine, and the future. The initiative responds to both optimism about AI’s potential and concerns about job loss, creative work, human agency, misuse, and whether benefits outweigh costs. Anthropic says its Public Benefit Corporation mission is to secure the benefits of advanced AI models and mitigate risks through safeguards, alignment research, free access for scientists, and a fellowship pairing early-career Claude users with nonprofits. It has gathered views through surveys of 52,000 Americans and 81,000 Claude users in 159 countries and 70 languages, plus focus groups and anonymized usage data, and will publicly track actions and shortcomings in addressing submitted questions. The Anthropic Institute and Long-Term Benefit Trust are also described as mechanisms for confronting societal challenges and providing oversight.
A Practical Guide to GPU-Initiated Communication for Molecular Dynamics at Scale
Molecular dynamics packages like GROMACS face severe scaling bottlenecks when CPU-orchestrated MPI communication introduces blocking synchronizations during halo exchanges. To resolve this, researchers implemented GPU-initiated communication using NVIDIA NVSHMEM to eliminate CPU-GPU handoffs. The redesigned approach fuses packing, remote memory access, and completion waiting into device-side kernels using fine-grained signaling. Benchmarks on NVIDIA DGX H100 systems demonstrated up to 1.5x performance gains intra-node and up to 2x on GB200 NVL72 multi-node clusters, particularly on latency-bound systems. While device-initiated communication introduces a slight 1–3% overhead on very large compute-bound configurations at low node counts, it consistently improves strong scaling for production workloads.
GitFarm: Git® as a Service for Large-Scale Monorepos
At Uber’s monorepo scale, traditional Git workflows create bottlenecks through multi-gigabyte clones, local checkouts, repeated synchronization, and independent fetch or push operations across automation systems. The post describes GitFarm, a Git as a Service platform exposed through a high-performance gRPC API that executes standard Git commands in secure, ephemeral sandboxes backed by pre-warmed repository and container pools. Each backend maintains a bare clone, receives push-based updates and periodic git fetch operations every five minutes, while clients can explicitly fetch when they need current state. The design provides a full checkout in under 500 milliseconds and reduced client-side resource utilization by over 80%, while avoiding local clones and long initial clone delays.
A new way to reflect on how you use Claude
The beta reflection feature helps Claude users review and refine their use of the assistant by tracking and visualizing activity over one, three, six, or twelve months. A dashboard summarizes key topics, usage patterns, frequent tasks, and when Claude is used, while periodically prompting users to consider what they should keep doing themselves; a future view will show time spent. It also assesses activity through the 4D AI Fluency Framework—Delegation, Description, Discernment, and Diligence—and can suggest practices such as using a Project for ongoing context, alongside quiet hours and break nudges. Available on Claude for web and the desktop app to Free, Pro, and Max users with memory enabled, the beta excludes incognito chats, connected-tool source files, and conversations tied to health integrations, with sensitive conversations represented only at a high level.
How Trigger.dev is using ClickHouse to scale observability for long-running AI workflows
Trigger.dev, an open-source platform for serverless asynchronous tasks, needed to scale observability for increasingly complex, long-running AI workflows spanning thousands of tenants. Its initial Postgres design stored logs, traces, and events in a growing table, making indexes, analytics, and operational reliability increasingly difficult to manage. The team moved observability workloads to ClickHouse after evaluating performance, scalability, open-source compatibility, deployment flexibility, and support for dynamic telemetry, including its native JSON data type. Migration occurred gradually at the task-execution level, routing each run’s logs to either database without downtime or customer intervention. ClickHouse now supports real-time telemetry analysis, approximately 200-millisecond P95 latency for queries returning tens of thousands of rows, and a dramatically smaller telemetry storage footprint.
amazon.science
Capturing token IDs during agentic interactions for better reinforcement learning
Reinforcement learning for agentic tasks requires updating model parameters against the precise token sequences generated during rollouts. Standard agent harnesses frequently modify conversation transcripts through message compaction and tool formatting, causing retokenization drift that degrades policy-gradient training. Turnstile addresses this issue as a Rust-based proxy situated between agent harnesses and inference backends such as SGLang. By intercepting OpenAI-compatible Chat Completions traffic, the proxy records exact token IDs, log probabilities, loss masks, weight versions, and mixture-of-experts routing traces at the moment of generation. Validation across text-based coding and multimodal computer-use agents demonstrated steady reinforcement learning improvements without requiring modifications to agent harnesses.
When and what should I be logging?
Targeted structured logging provides a fast way to capture operational behavior in production without requiring new deployments for debugging. Developers benefit most from logging key runtime decisions, multi-step algorithm outcomes, mutating audit operations, and context around non-critical or retried errors. Rather than outputting plain text, applications should record structured key-value pairs that capture who performed an action, what occurred, and associated trace identifiers. Sensitive information such as secrets and personally identifiable data must be excluded or scrubbed, and large unparsed payloads should be avoided to prevent excessive costs. Managing log levels appropriately and treating logs as potentially temporary instrumentation helps maintain actionable telemetry without generating unnecessary noise.
Searchable field-level encryption on Supabase with CipherStash
CipherStash’s integration for Supabase adds searchable field-level encryption to Postgres applications without requiring a schema change. The encrypted Supabase SDK wrapper encrypts marked fields in the application, stores each value as a JSON payload containing ciphertext and Searchable Encrypted Metadata (SEM), and converts queries to SEM so Postgres can filter, sort, and join without plaintext access. ZeroKMS keeps keys under the customer’s control, while decryption policies determine who can read each value; the source states that neither CipherStash nor Supabase can access plaintext data. The integration supports Supabase.js, Drizzle, and Prisma Next, with a one-command setup, and CipherStash Proxy covers analytics, background workers, and other direct database access through the Postgres wire protocol.
Top topics
- Architecture834
- Performance816
- LLMs660
- Machine Learning622
- Open Source563
- AI Agents553
- AI470
- Developer Experience448
Weekly digest
The best engineering reads, once a week.