Tun
@stereosky.bsky.social
📊 Data wrangler
🎸 Guitar riffer
🏄♂️ Surf dreamer
🍦 Ice cream eater
🗣 Opinions are m̶y̶ ̶o̶w̶n̶ open source
🐍 #PyData Organiser @pydatacornwall.bsky.social
@pydatalondon.bsky.social
🎸 Guitar riffer
🏄♂️ Surf dreamer
🍦 Ice cream eater
🗣 Opinions are m̶y̶ ̶o̶w̶n̶ open source
🐍 #PyData Organiser @pydatacornwall.bsky.social
@pydatalondon.bsky.social
😎 The team also had the great idea of keeping the prizes (Meta Ray-Bans) in full view so the hackers remember it's a competition with huge stakes. Looking forward to judging the winners at the end of the day
October 15, 2025 at 4:33 PM
😎 The team also had the great idea of keeping the prizes (Meta Ray-Bans) in full view so the hackers remember it's a competition with huge stakes. Looking forward to judging the winners at the end of the day
🇺🇸 Got to hand it to the design team for these stickers with our favourite American classics (and a reminder to use the Lenses MCP server — launching soon).
October 15, 2025 at 4:33 PM
🇺🇸 Got to hand it to the design team for these stickers with our favourite American classics (and a reminder to use the Lenses MCP server — launching soon).
🗽 We are live at the Real-Time Data AI Hackathon with Lenses.io x Amazon Web Services (AWS) x Ness Digital Engineering, enjoying the creativity and views.
October 15, 2025 at 4:33 PM
🗽 We are live at the Real-Time Data AI Hackathon with Lenses.io x Amazon Web Services (AWS) x Ness Digital Engineering, enjoying the creativity and views.
Nick Aldrich from @awscloud.bsky.social showing us what's next for async tool calls in MCP (though I think an event-driven agent approach with ephemeral clients is the way). There will be improvements in client-polling. Follow and contribute to it in SEP-1391 github.com/modelcontext... #MCPDevSummit
October 2, 2025 at 12:42 PM
Nick Aldrich from @awscloud.bsky.social showing us what's next for async tool calls in MCP (though I think an event-driven agent approach with ephemeral clients is the way). There will be improvements in client-polling. Follow and contribute to it in SEP-1391 github.com/modelcontext... #MCPDevSummit
Full house at the #MCPDevSummit morning keynotes. Great seeing David Soria Parra from @anthropic.com telling us the story so far: 10,000+ integrations built by the community in the 10 months since the MCP launch.
October 2, 2025 at 12:34 PM
Full house at the #MCPDevSummit morning keynotes. Great seeing David Soria Parra from @anthropic.com telling us the story so far: 10,000+ integrations built by the community in the 10 months since the MCP launch.
🫡 I salute @workos.bsky.social for providing secure auth for MCP servers and also their commitment to swag at #MCPDevSummit
October 2, 2025 at 12:33 PM
🫡 I salute @workos.bsky.social for providing secure auth for MCP servers and also their commitment to swag at #MCPDevSummit
I'm heading to the MCP Developers Summit tomorrow. Who else will I see there?
Really looking forward to the keynotes, spec updates (OAuth in particular) and war stories from MCP in production. Also quietly hoping for swag of this magnitude 🤞🏽 @mcpdevsummit.ai
Really looking forward to the keynotes, spec updates (OAuth in particular) and war stories from MCP in production. Also quietly hoping for swag of this magnitude 🤞🏽 @mcpdevsummit.ai
October 1, 2025 at 9:02 AM
I'm heading to the MCP Developers Summit tomorrow. Who else will I see there?
Really looking forward to the keynotes, spec updates (OAuth in particular) and war stories from MCP in production. Also quietly hoping for swag of this magnitude 🤞🏽 @mcpdevsummit.ai
Really looking forward to the keynotes, spec updates (OAuth in particular) and war stories from MCP in production. Also quietly hoping for swag of this magnitude 🤞🏽 @mcpdevsummit.ai
Enjoyed a lovely pint on Roger Daltrey’s estate (it absolutely pissed it down 2 mins later)
August 30, 2025 at 3:53 PM
Enjoyed a lovely pint on Roger Daltrey’s estate (it absolutely pissed it down 2 mins later)
It’s a marathon, not a sprint.
This is a well known phrase, especially if you’re a founder or if you work in a startup. It absolutely is a marathon and not a sprint, but our concept of a good finishing time is broad.
This is a well known phrase, especially if you’re a founder or if you work in a startup. It absolutely is a marathon and not a sprint, but our concept of a good finishing time is broad.
August 30, 2025 at 9:59 AM
It’s a marathon, not a sprint.
This is a well known phrase, especially if you’re a founder or if you work in a startup. It absolutely is a marathon and not a sprint, but our concept of a good finishing time is broad.
This is a well known phrase, especially if you’re a founder or if you work in a startup. It absolutely is a marathon and not a sprint, but our concept of a good finishing time is broad.
"add a tool that makes me a sandwich" is the new "sudo make me a sandwich".
Here's my take on the Sandwich kxcd. My AI-coding tool of choice is Cline. My favourite feature: it can extend its own capabilities via Model Context Protocol. If it can't do something, you can just ask it to add a tool.
Here's my take on the Sandwich kxcd. My AI-coding tool of choice is Cline. My favourite feature: it can extend its own capabilities via Model Context Protocol. If it can't do something, you can just ask it to add a tool.
July 22, 2025 at 11:13 AM
"add a tool that makes me a sandwich" is the new "sudo make me a sandwich".
Here's my take on the Sandwich kxcd. My AI-coding tool of choice is Cline. My favourite feature: it can extend its own capabilities via Model Context Protocol. If it can't do something, you can just ask it to add a tool.
Here's my take on the Sandwich kxcd. My AI-coding tool of choice is Cline. My favourite feature: it can extend its own capabilities via Model Context Protocol. If it can't do something, you can just ask it to add a tool.
Vibe coding is like driving with GPS.
You still need to pay attention to motorists, pedestrians and road safety. If you overuse it, you lose memory of routes you know. If you don't provide details and accept what you're given, you may end up in Kansas City, Kansas instead of Kansas City, Missouri
You still need to pay attention to motorists, pedestrians and road safety. If you overuse it, you lose memory of routes you know. If you don't provide details and accept what you're given, you may end up in Kansas City, Kansas instead of Kansas City, Missouri
July 21, 2025 at 6:26 PM
Vibe coding is like driving with GPS.
You still need to pay attention to motorists, pedestrians and road safety. If you overuse it, you lose memory of routes you know. If you don't provide details and accept what you're given, you may end up in Kansas City, Kansas instead of Kansas City, Missouri
You still need to pay attention to motorists, pedestrians and road safety. If you overuse it, you lose memory of routes you know. If you don't provide details and accept what you're given, you may end up in Kansas City, Kansas instead of Kansas City, Missouri
I recently realised I refer to "Small LLMs" and "Large LLMs". This may seem like RAS (Redundant Acronym Syndrome) Syndrome, e.g. PIN Number and ATM Machine but it isn't. To me...
July 18, 2025 at 10:56 AM
I recently realised I refer to "Small LLMs" and "Large LLMs". This may seem like RAS (Redundant Acronym Syndrome) Syndrome, e.g. PIN Number and ATM Machine but it isn't. To me...
July 16, 2025 at 9:36 AM
60% of the time, it’s salami every time
June 25, 2025 at 11:50 PM
60% of the time, it’s salami every time
The fins are back on the desk so it's The Fin Room once again. Love the stubby Almond Surfboards one and the pink Watermark Surfboards one too
June 25, 2025 at 1:30 PM
The fins are back on the desk so it's The Fin Room once again. Love the stubby Almond Surfboards one and the pink Watermark Surfboards one too
🧘♂️ Must remember to take more breaks from the screen and stare into the horizon. Lucky to have this view ~2 mins walk away from today's desk
June 25, 2025 at 12:33 PM
🧘♂️ Must remember to take more breaks from the screen and stare into the horizon. Lucky to have this view ~2 mins walk away from today's desk
Today's the day of the first PyData Cornwall meetup! Prepped the event space this morning so it's all clean and tidy. Someone mentioned this is the most expensive cleaning team (comprised of data engineers), but there's no job too small when it comes to building community
@pydatacornwall.bsky.social
@pydatacornwall.bsky.social
June 25, 2025 at 11:02 AM
Today's the day of the first PyData Cornwall meetup! Prepped the event space this morning so it's all clean and tidy. Someone mentioned this is the most expensive cleaning team (comprised of data engineers), but there's no job too small when it comes to building community
@pydatacornwall.bsky.social
@pydatacornwall.bsky.social
Delighted with these scones being Cornish rather than Devonshire #IYKYK #CreamOnTop
June 8, 2025 at 1:52 PM
Delighted with these scones being Cornish rather than Devonshire #IYKYK #CreamOnTop
Final #PyData London keynote from Tony Mears on the ways that innovation dies, from his experience directing operations in the NHS, UK Space Agency and government agencies. Great insights into how innovation gets stifled by risk aversion and lack of a failure culture
@pydatalondon.bsky.social
@pydatalondon.bsky.social
June 8, 2025 at 1:05 PM
Final #PyData London keynote from Tony Mears on the ways that innovation dies, from his experience directing operations in the NHS, UK Space Agency and government agencies. Great insights into how innovation gets stifled by risk aversion and lack of a failure culture
@pydatalondon.bsky.social
@pydatalondon.bsky.social
A comprehensive dive into why you should care about LLM evaluations from Lena Shakurova at #PyData London. Build datasets with SMEs. When you have an understanding then create synthetic data. Good tips for metrics and creating a 3 step evals framework
@pydatalondon.bsky.social
@pydatalondon.bsky.social
June 8, 2025 at 10:39 AM
A comprehensive dive into why you should care about LLM evaluations from Lena Shakurova at #PyData London. Build datasets with SMEs. When you have an understanding then create synthetic data. Good tips for metrics and creating a 3 step evals framework
@pydatalondon.bsky.social
@pydatalondon.bsky.social
Some highlights: Using LLMs to figure out if PyData talks have become repetitive over the years, Tarot readings with a mobile printer and an LLM (to great success!) and @hevansdev.bsky.social explaining his love/hate relationship with meetup platforms
June 8, 2025 at 9:38 AM
Some highlights: Using LLMs to figure out if PyData talks have become repetitive over the years, Tarot readings with a mobile printer and an LLM (to great success!) and @hevansdev.bsky.social explaining his love/hate relationship with meetup platforms
⚡️ Lightning talks this morning at #PyData London. Loads of fun submissions; the wheel decides
@pydatalondon.bsky.social
@pydatalondon.bsky.social
June 8, 2025 at 8:05 AM
⚡️ Lightning talks this morning at #PyData London. Loads of fun submissions; the wheel decides
@pydatalondon.bsky.social
@pydatalondon.bsky.social
Going deep into transformer layers with @baggiponte.bsky.social explaining how to count parameters in order to calculate compute and memory requirements. Note to self: get an understanding of key-value (KV) caching in transformers
#PyData London
@pydatalondon.bsky.social
#PyData London
@pydatalondon.bsky.social
June 7, 2025 at 3:56 PM
Going deep into transformer layers with @baggiponte.bsky.social explaining how to count parameters in order to calculate compute and memory requirements. Note to self: get an understanding of key-value (KV) caching in transformers
#PyData London
@pydatalondon.bsky.social
#PyData London
@pydatalondon.bsky.social
Zines being made in the Feminist AI Lounge hosted by @inesmontani.bsky.social - some amazing stickers making their way onto collages from some creative geniuses at #PyData London
@feministai.party @pydatalondon.bsky.social
@feministai.party @pydatalondon.bsky.social
June 7, 2025 at 3:18 PM
Zines being made in the Feminist AI Lounge hosted by @inesmontani.bsky.social - some amazing stickers making their way onto collages from some creative geniuses at #PyData London
@feministai.party @pydatalondon.bsky.social
@feministai.party @pydatalondon.bsky.social
Adding GenAI to an existing working ML system seems unnecessary but @astroadamh.bsky.social goes through reasons to do it whilst navigating pitfalls, e.g. changes in model prices. Great idea exploring this through personas we’ve all encountered in the workplace
#PyData @pydatalondon.bsky.social
#PyData @pydatalondon.bsky.social
June 7, 2025 at 2:59 PM
Adding GenAI to an existing working ML system seems unnecessary but @astroadamh.bsky.social goes through reasons to do it whilst navigating pitfalls, e.g. changes in model prices. Great idea exploring this through personas we’ve all encountered in the workplace
#PyData @pydatalondon.bsky.social
#PyData @pydatalondon.bsky.social