Diogo Santos’ Post

PSA for devs using HuggingFace Inference API for embeddings Spent some time debugging a cryptic error today and wanted to share the fix. If you're using router.huggingface.co with sentence-transformers/all-MiniLM-L6-v2, you might hit this: https://lnkd.in/dAcz8rmk() missing 1 required positional argument: 'sentences' Here's what's happening: The HF inference router determines the pipeline from the model's tags. all-MiniLM-L6-v2 gets routed to the SentenceSimilarity pipeline instead of feature-extraction — so the request payload format doesn't match what the pipeline expects. The fix? Use a model HuggingFace explicitly tags as feature-extraction. I switched to BAAI/bge-base-en-v1.5 and the error went away immediately. A few more things worth knowing: → https://lnkd.in/dxdEg7gX is now deprecated and returns 410 Gone. You can't fall back to the old endpoint — use router.huggingface.co going forward. → all-MiniLM-L6-v2 outputs 384-dimensional vectors. If your pgvector column is defined as vector(768), you have a silent dimension mismatch you may not have caught yet. BAAI/bge-base-en-v1.5 outputs 768 dims, so switching also fixed that for us. TL;DR: 1. Check that your model is tagged for the right pipeline on HuggingFace 2. Make sure your embedding dimensions match your vector column 3. https://lnkd.in/dxdEg7gX is gone — use router.huggingface.co Hope this saves someone a few hours. #softwaredevelopment #machinelearning #embeddings #huggingface #postgresql #pgvector #python #developers

🚀 FlameIQ v1.0.2 Released FlameIQ is an open-source performance regression detection engine designed for CI environments. The tool compares benchmark results against a stored baseline on every CI run and detects regressions using configurable thresholds and optional statistical testing. Key capabilities • Compares benchmark results against a stored baseline on every CI run • Enforces per-metric thresholds with direction-aware regression logic • Optional Mann–Whitney U statistical significance testing • Generates self-contained HTML performance reports • Outputs machine-readable JSON results for CI pipelines Installation pip install flameiq-core Resources Documentation: https://lnkd.in/d6e2D7mq PyPI: https://lnkd.in/d-2KcKFd Source Code: https://lnkd.in/d2VDWRQa Contributions and feedback are welcome as the project continues to evolve. #opensource #performanceengineering #python #devtools #cicd

This article shows how to add dynamic voice alerts to MetaTrader 5 EAs using a local text-to-speech pipeline, avoiding the limitations of pre-recorded WAV notifications. An MQL5 Expert Advisor formats market events (signals, prices, time, ATR, margin/trade status) into text and sends it via WebRequest HTTP POST to a localhost Python service. The Python side exposes a simple /speak endpoint, queues incoming messages, and speaks them sequentially to prevent overlaps and thread-related crashes. A moving-average crossover EA demonstrates the pattern, including a TestMode that speaks on each new bar and practical error handling for common WebRequest failures. This design improves accessibility, reduces alert maintenance, and stays fully local for low latency and privacy. #MQL5 #MT5 #EA #TTS https://lnkd.in/dRx_7Qze

I wrote a CLI tool to help your teams catch "slop" code before it creates a mountain of tech debt. Runs for any python projects 3.10 -> 3.13. Drop it in your CI/CD pipelines to act as a last linter for AI specific slop patterns like vague comments, util functions, and more. Check it out! https://lnkd.in/e5xNUwH2 https://lnkd.in/eT6psfZS

🚀 Debugging Journey: Largest BST in a Binary Tree (with Solution) Today I worked on a classic DSA problem — finding the largest BST inside a Binary Tree — and it really tested my debugging + recursion skills. 💡 Key Learnings: 🔹 Use correct boundary values → float('inf'), float('-inf') 🔹 Always track multiple things in recursion (size, min, max, BST status) 🔹 Correct condition is: max(left) < root < min(right) 🔹 Small syntax mistakes can break the whole logic ✅ Approach: For every node, return: Size of BST Minimum value Maximum value Whether it's a BST If subtree is BST → combine left + right Else → take max of left/right subtree 💻 Python Solution: class Solution: def largestBst(self, root): def helper(node): if not node: return 0, float('inf'), float('-inf'), True N1, min1, max1, isBST1 = helper(node.left) N2, min2, max2, isBST2 = helper(node.right) if isBST1 and isBST2 and max1 < node.data < min2: return (N1 + N2 + 1, min(min1, node.data), max(max2, node.data), True) return max(N1, N2), 0, 0, False ans, _, _, _ = helper(root) return ans 📌 Key Takeaway: “Debugging isn’t just fixing code — it’s understanding your logic deeply.” This problem helped me improve: ✔️ Tree Traversal ✔️ Recursion Thinking ✔️ Debugging Mindset #DSA #Python #BinaryTree #CodingJourney #Debugging #LearnInPublic #TechSkills

Secrets Hunter v0.6.0 is here! Most secret scanners ask one question: "Does this suspicious string match a known pattern?" Secrets Hunter asks a better one: "What is the context this suspicious string was found in?" A high-entropy string alone means little. But a high-entropy string next to API_KEY= is a different story entirely. Secrets Hunter combines Shannon entropy with assignment context analysis and a bigram language model to catch secrets that no pattern library would ever cover — including custom tokens, internal credentials, and formats unique to your stack. Here's what makes it different: → Assignment context boosting — severity is determined not just by the string itself, but by what it's assigned to → Semantic false positive filtering — strings that look like human language get filtered out before they reach you → Zero dependencies — pure Python 3.11+, no binaries, no system packages, no network calls at scan time → TOML overlay config — stack team, CI, and local configs without duplicating anything → SARIF + JSON export — plugs directly into GitHub Code Scanning and any CI pipeline Built from real AppSec experience. Lightweight by design. Fully auditable. 👉 pip install secrets-hunter Visit Secrets Hunter repository: https://lnkd.in/dYNysfgW #AppSec #DevSecOps #SecretScanning #Python #OpenSource #SecurityTools

LinkedIn: 📣 SynapseKit v0.6.9 is live. Two graph features in this release that I think matter more than they look. approval_node(): gates your graph on a human decision. The workflow hits a node, pauses, waits for a human to approve or reject, then continues. No polling, no hacks. One function call. dynamic_route_node(): routes to completely different subgraphs at runtime based on whatever logic you write. Sync or async. Your graph decides where it goes next while it's running. Together these two make human-in-the-loop workflows actually practical to build. Not a demo. Production. Also shipped: 💬 SlackTool [Slack]— send messages via webhook or bot token 📋 JiraTool— search, create, comment on issues via REST 🔍 BraveSearchTool [Brave]— web search via Brave API All three stdlib only. Zero new dependencies. Where we stand: 32 tools · 15 providers · 18 retrieval strategies · 795 tests · 2 dependencies. ⚡ pip install synapsekit 🔗 https://lnkd.in/d2fGSPkX #Python #LLM #RAG #OpenSource #AI #MachineLearning #Agents #SynapseKit

I thought contributing to Streamlit meant 𝗳𝗶𝘅𝗶𝗻𝗴 𝘀𝗺𝗮𝗹𝗹 𝗨𝗜 𝗯𝘂𝗴𝘀. I didn’t expect to 𝗿𝗲𝘃𝗲𝗿𝘀𝗲-𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿 its entire architecture. While working on a component, I kept asking: “𝗪𝗵𝗲𝗿𝗲 𝗱𝗼𝗲𝘀 𝘁𝗵𝗶𝘀 𝗯𝘂𝘁𝘁𝗼𝗻 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗴𝗼?” I became so curious that I kept learning more and more about it 👇 User Python Code ↓ Elements API (button.py) ↓ DeltaGenerator ↓ Protobuf ↓ Runtime / Session Manager ↓ Tornado WebSocket ↓ React (TSX) ↓ Browser UI Unlike 𝗗𝗷𝗮𝗻𝗴𝗼 𝘁𝗲𝗺𝗽𝗹𝗮𝘁𝗲𝘀 or 𝗦𝗽𝗿𝗶𝗻𝗴 𝗕𝗼𝗼𝘁 𝗥𝗘𝗦𝗧 𝗔𝗣𝗜𝘀, Streamlit flows like this: 🧠 Python → Protobuf → WebSocket → React → Rerun Model Every click reruns the script. State lives in the session manager. UI updates happen through delta patches over a persistent WebSocket. That’s when it clicked, Streamlit isn’t just client–server. It’s reactive execution architecture, closer to 𝗥𝗲𝗮𝗰𝘁 + 𝗝𝘂𝗽𝘆𝘁𝗲𝗿 + 𝗲𝘃𝗲𝗻𝘁-𝗱𝗿𝗶𝘃𝗲𝗻 systems. The lesson? If a framework feels “𝘀𝗶𝗺𝗽𝗹𝗲,” it’s usually hiding something sophisticated. Don’t stop at using it - dissect it. Understanding the layers transforms you from a user into a builder. How do you usually break down a complex system to understand it completely? #CodingJourney #DeveloperCommunity #Streamlit #OpenSourceContributing

Claude can write a compiler, but it can't spell "anthropic" backwards. I tested it: pure LLM: "cipohtrpna" => wrong. Give it a sandbox? It writes `echo "anthropic" | rev`, runs it, gets "ciporhtna." => correct 10 tests where LLMs fail: counting, reversal, prime factorization, set ops. No sandbox: 8/10. With sandbox: 10/10. I didn't tell it to write code, it just recognized what it's bad at and reached for Python! Anthropic just shipped Programmatic Tool Calling (Claude writes code that calls your tools). Fewer round-trips, less tokens. It runs in their container, not yours. Testing it, something occurred to me: does that still work without tools? Yes, and it patches its own blind spots with code! Have you tried PTC already? #AICoding #ClaudeCode #AIAgents #LLM

Everyone learns stacks. But very few understand where they actually matter. Take a simple problem: Checking if brackets are balanced. Most people think it’s about counting. It’s not. It’s about order. Here’s what really happens behind the scenes: → You scan the expression left to right → Every opening bracket goes into a stack → Every closing bracket tries to match the last opening one If it matches → remove it If it doesn’t → the entire structure breaks That’s the moment you realize: Stacks aren’t just data structures. They are decision systems. They enforce rules like: Last In → First Out And that’s exactly how: • Code editors validate syntax • Compilers detect errors • Browsers manage navigation history A simple example: [(a+b)] → Valid ✔ [(a+b] → Invalid ❌ Same characters. Different structure. That’s the difference between working code and broken logic. The lesson? In programming — and in systems — structure beats quantity. Always. #DataStructures #Python #ProblemSolving #CodingJourney #AIThinking