|
| 1 | +Date: 30-08-2025 |
| 2 | + |
| 3 | +Duration: 3 hours,4 minutes |
| 4 | + |
| 5 | +==== Topics Discussed |
| 6 | + |
| 7 | +* Ayush kicked off a discussion on probability and algorithms, specifically diving into entropy encoders like Huffman, Arithmetic, rANS, and tANS. |
| 8 | + ** He broke down concepts like `Log2(number of symbols)` and `2^n` in coding. |
| 9 | + ** He also explained arithmetic coding, highlighting its efficiency from `O(2^n)` to `O(n)`, and touched on cumulative frequency and probability functions. |
| 10 | + *** link:https://stanforddatacompressionclass.github.io/Fall22/lectures[Stanford Data Compression Class Lectures^] |
| 11 | +* Lucky Mee, who's returning to tech after some time, asked for advice. |
| 12 | + ** Harsh strongly suggested focusing on system designs and also emphasized the importance of planning thoroughly before execution. |
| 13 | +* Abhigyan showcased his project on stock analysis using AI agents. |
| 14 | + ** He shared a research report he generated from the agents and demonstrated how these reports provided structured outputs, detailing weaknesses, strengths, conclusions, graphs, and metrics. |
| 15 | + *** link:https://www.deeplearning.ai/the-batch[DeepLearning.AI The Batch Newsletter^] |
| 16 | + *** link:https://platform.openai.com/docs/guides/structured-outputs[OpenAI Guides: Structured Outputs^] |
| 17 | +* Chetan Jahangir shared some insights about Builder AI. |
| 18 | + ** He mentioned how the company allegedly scammed Microsoft. |
| 19 | + ** This sparked a conversation about how we should approach data derived from AI and LLMs. |
| 20 | +* Ayush provided an update on his SIH project. |
| 21 | + ** He's concentrating on smart system design for hardware. |
| 22 | + ** He's currently learning about the STM32 microprocessor. |
| 23 | + *** link:https://en.wikipedia.org/wiki/STM32[STM32 Wikipedia^] |
| 24 | + *** link:https://www.st.com/en/evaluation-tools/stm32-nucleo-boards.html[STM32 Nucleo Boards - STMicroelectronics^] |
| 25 | +* Harsh questioned how MCP (Multiple Connected Models) relates to agentic AI. |
| 26 | + ** Abhigyan and Chetan clarified that it's essentially multiple models linked to a central one. |
| 27 | + ** They illustrated this with an example: a request for a weather update goes to the MCP, which then routes it to various models, with the configuration depending on the user. |
| 28 | +* Chetan Jahangir sought advice on getting started with research papers. |
| 29 | + ** He asked for tips on how to begin and how to connect with people who write research papers to get involved and share knowledge. |
| 30 | + ** Rishit suggested reaching out to seniors and their professors. |
| 31 | + *** link:https://icml.cc/Conferences/2022/BestPractices[ICML 2022 Best Practices for Research^] |
| 32 | +* Swapnil Borkar discussed AI and its impact on jobs. |
| 33 | + ** He shared concerns about how AI could threaten junior roles if not utilized effectively. |
| 34 | +* Jaden Furtado talked about UPI, UPI Lite, and UPI Lite X. |
| 35 | + ** link:https://www.npci.org.in/what-we-do/upi-lite/product-overview[UPI Lite Product Overview - NPCI^] |
| 36 | + ** link:https://www.npci.org.in/what-we-do/upi-lite/upi-lite-x/product-overview[UPI Lite X Product Overview - NPCI^] |
| 37 | +* Additional resources and links |
| 38 | + ** link:https://people.math.harvard.edu/~ctm/home/text/others/shannon/entropy/entropy.pdf[A Mathematical Theory of Communication by C. E. Shannon^] |
| 39 | + ** link:https://stanforddatacompressionclass.github.io/Fall22[Stanford Data Compression Course Home^] |
| 40 | + ** link:https://youtu.be/BiuknvAYDPs[OTC Talks #7: The Browser Backstage - Peek-in-Perf by @VivekGalatage^] |
| 41 | + ** link:https://youtu.be/8_GgeASwHwQ[Google’s nano banana just killed Photoshop... let’s run it - Fireship^] |
| 42 | + ** link:https://youtu.be/QgVR1dxy6zk[it only took 2 lines of code... - Low Level^] |
| 43 | + ** link:https://youtu.be/fpkCNu_dzcM[CPU Summary - Computerphile^] |
| 44 | + ** link:https://youtu.be/tnPAhVxsPHE[Your iPhone is Lying to You About Files... - LaurieWired^] |
| 45 | + ** link:https://youtu.be/D4hNLgJmshg[The Operating System Known As GNU/Linux - Brodie Robertson^] |
| 46 | + ** link:https://youtu.be/O9QD9an-0jU[well this isn't great... - Low Level^] |
| 47 | + ** link:https://youtu.be/pOF11EDprxc[Deno vs Oracle: The ugly custody battle for JavaScript… - Fireship^] |
| 48 | + ** link:https://youtu.be/rcBVqcoQkuA([Bash Pipes in C - Tsoding Daily^] |
| 49 | + ** link:https://youtu.be/_bbuRFT2l-Q[AI Trends in 2025 - Gaurav Sen^] |
| 50 | + ** link:https://youtu.be/2JzOe1Hs26Q[Accidentally Built a Nuclear Supercomputer. - PewDiePie^] |
| 51 | + ** link:https://youtu.be/fzlflyw7X2I[The Next Big SHA? SHA3 Sponge Function Explained - Computerphile^] |
| 52 | + ** link:https://youtu.be/55MwAlcpvK4[The Operating System Known As Linux - Brodie Robertson^] |
| 53 | + ** link:https://ai.pydantic.dev/agents[Pydantic AI Agents Documentation^] |
| 54 | + ** link:https://youtu.be/PhCrlpUwEhU[Agentic RAG: An End-to-End Open Source Framework - Weaviate^] |
| 55 | + ** link:https://medium.com/@thomas_78526/builder-ai-was-a-scam-d1f8fdd16c3f[Builder.ai Was a Scam - Medium Article^] |
| 56 | + ** link:https://github.com/TecharoHQ/anubis[Anubis GitHub Project (TecharoHQ)^] |
| 57 | + ** link:https://www.beatflow.art[Beatflow Art Platform^] |
0 commit comments