Ask HN: What are you working on (August 2024)?

Hacker News Discussion ↗

AI and ML Projects

  • Many are building LLM-powered tools: on‑call copilots, desktop/QA agents that control UI, “autopilot” for OSes, RAG → Graph RAG upgrades, multi‑agent cognitive architectures, and SMS/phone “firewalls” using small models on-device.
  • Several focus on information extraction and summarization: email/download/search tools, podcast/audio analysis platforms, “AI newspaper” and news aggregators, video/audio → text “ChatGPT for video”, and site chatbots.
  • Others explore deeper research: graph reasoning, agent ethics layers (e.g. Stoic-inspired), representation learning (V‑JEPA on chess), reinforcement learning beyond Gym, and ML-on-microcontrollers (MicroPython).

Developer Tools, Infra, and Data

  • New CI/CD and build systems emphasize local-first runs, DAG definitions, caching, and easier debugging; some compare to Bazel and Dagger and worry about scaling and “abstraction ceilings”.
  • Numerous libraries: LLM-as-function call wrappers, property-testing for TypeScript, Bun-based stacks, serialization formats, open-source OAuth2, CLI frameworks, Nim hot‑reloading, multi-cloud observability pipelines, and “virus‑scanner style” deepfake detection.
  • Infra tooling includes: k8s-style protocols for complex RPC, GUIs for Postgres on mobile, Git-based task and issue tracking, OpenTelemetry-like observability for traditional environments, and tools to detect relicensing risk and CLAs in OSS.
  • Several build personal or hybrid data stores: out-of-memory dataframes, personal knowledge bases spanning email/calendar/files, and self-hosted encrypted password+key directories for the fediverse.

Apps, SaaS, and Productivity

  • Many small SaaS: incident management, API monetization and billing, self‑service CI, lending platforms, payroll for specific verticals, error debuggers, AI job‑application agents, screening tools for applicants, and pricing/lead scrapers.
  • Productivity/PKM is a strong theme: note-taking and journaling apps, blogging-by-email, RSS readers, email downloaders with search and labeling, to‑do/“text yourself” tools, and digital gardens.
  • Niche consumer apps include: real estate search with commute/park filters, city‑walk/tour generators, speech‑to‑speech translation, language learning games, fashion/style analyzers, microfinance-style bill consolidators, and local event finders.

Games, Creative, and Hardware

  • Indie games and engines: strategy and roguelike titles, NES emulators and engines, deck‑builders, automation sandboxes, VR concepts, plus tooling for table‑top RPGs and backgammon.
  • Creative tools span: CSS masterclasses, UI/UX courses, audio players, shader/SDF playgrounds, 3D art systems, and platforms for podcasts and interactive video.
  • Hardware and physical builds include: EV batteries and right‑to‑repair projects, IoT meters, CAN bus reverse‑engineering, drones and simulators, CNC/lithography experiments, fingerboards, 3D‑printed braille cells, and home labs.

Learning, Careers, and Personal Life

  • Many are learning new stacks (Astro, SwiftUI, Unity, RL, materials science, Go, Rust, gnuplot) or writing books and tutorials (Git, coding edtech, AI agents, CSS, Lisp).
  • Several posts focus on mental and physical health, burnout, therapy, and rehab; commenters emphasize exercise, PT, and self‑care as legitimate “projects” alongside tech work.