Folder: Notes/Tech_Journal/2024

Dec 31, 2024

fixing sleep wake hangs on linux with amd gpus as a boundary test

Dec 31, 2024

may thy bits chip and shatter patterns building observability pipelines at scale as a trust problem

Dec 31, 2024

microsoft's ai setback an imaginary letter as a trust problem

Dec 31, 2024

reading closed failing charter reopened as private religious school funded by vouchers as a constraint shift

Dec 31, 2024

reading s&p 500 posts 23% gain for 2024 as a constraint shift

Dec 31, 2024

the part of positive emotions from brain injury the emergence of mirth and happiness that changes behavior

Dec 31, 2024

the sharp edge behind deep learning gpu selector

Dec 31, 2024

the sharp edge behind ibm will release the largest ever quantum computer in 2025

Dec 31, 2024

timeline of ai models released in 2024 and the cost of defaults

Dec 30, 2024

gpu glossary as a boundary test

Dec 30, 2024

ts_server a web server proposing a rest api to large language models as a boundary test

Dec 24, 2024

token budgeting middleware prevents runaway agent loops

Dec 23, 2024

deterministic tool mocks accelerate agent test cycles

Dec 22, 2024

queue aware batching improves gpu utilization stability

Dec 21, 2024

eval driven deployment gates reduce regression churn

Dec 08, 2024

vector freshness daemons improve retrieval trustworthiness

Dec 07, 2024

context window compression pipelines lower serving spend

Dec 06, 2024

replay based debugging becomes standard for agent incidents

Dec 05, 2024

typed tool registries improve agent planner reliability

Dec 04, 2024

prompt cache invalidation strategies reduce tail latency

Dec 03, 2024

postgres logical replication feeds realtime rag refresh

Dec 02, 2024

wasm sandboxes cut plugin blast radius in agents

Dec 01, 2024

grpc gateway patterns stabilize tool calling backends

Nov 21, 2024

small language models win on edge maintenance

Nov 20, 2024

structured output contracts reduce agent failure rates

Nov 11, 2024

stateful agents gain safer rollback controls

Nov 08, 2024

open telemetry for llm traces matures

Oct 25, 2024

agentic observability stacks become standard

Oct 21, 2024

fast inference compilers close p95 gaps

Oct 15, 2024

cuda alternatives gain real benchmark traction

Oct 08, 2024

cloudflare workers ai adds regional controls

Oct 07, 2024

postgres vector indexing reaches mainstream ops

Oct 03, 2024

mistral large refresh targets enterprise latency

May 07, 2024

apple m4 neural engine shifts ondevice economics

Apr 18, 2024

Keith Kitchen

Explorer

Folder: Notes/Tech_Journal/2024

fixing sleep wake hangs on linux with amd gpus as a boundary test

may thy bits chip and shatter patterns building observability pipelines at scale as a trust problem

microsoft's ai setback an imaginary letter as a trust problem

reading closed failing charter reopened as private religious school funded by vouchers as a constraint shift

reading s&p 500 posts 23% gain for 2024 as a constraint shift

the part of positive emotions from brain injury the emergence of mirth and happiness that changes behavior

the sharp edge behind deep learning gpu selector

the sharp edge behind ibm will release the largest ever quantum computer in 2025

timeline of ai models released in 2024 and the cost of defaults

gpu glossary as a boundary test

ts_server a web server proposing a rest api to large language models as a boundary test

token budgeting middleware prevents runaway agent loops

deterministic tool mocks accelerate agent test cycles

queue aware batching improves gpu utilization stability

eval driven deployment gates reduce regression churn

vector freshness daemons improve retrieval trustworthiness

context window compression pipelines lower serving spend

replay based debugging becomes standard for agent incidents

typed tool registries improve agent planner reliability

prompt cache invalidation strategies reduce tail latency

postgres logical replication feeds realtime rag refresh

wasm sandboxes cut plugin blast radius in agents

grpc gateway patterns stabilize tool calling backends

small language models win on edge maintenance

structured output contracts reduce agent failure rates

stateful agents gain safer rollback controls

open telemetry for llm traces matures

agentic observability stacks become standard

fast inference compilers close p95 gaps

cuda alternatives gain real benchmark traction

cloudflare workers ai adds regional controls

postgres vector indexing reaches mainstream ops

mistral large refresh targets enterprise latency

apple m4 neural engine shifts ondevice economics

llama three launch pressures api only stacks

Stacked notes