talk-data.com

🎄 Build & Learn: Christmas Special

Week 6 — Wrap-Up, Polish & Presentation Readiness

Week 6 is about closing loops before Presentation Week (Week 7). By now, you’ve likely built something — an agent, a flow, a prototype, or even just a rough idea that works part of the time. This session is about turning that into a coherent story you can confidently show, explain, and reflect on. 💡 This is not about adding more features. It’s about stabilizing, simplifying, and making sense of what you already have. You can join anytime — no pressure to be “done.”

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Saturday Dec 13, 10:00 AM- 12 PM 📍 Where? : Joachimsthaler Str. 43, 10623 Berlin ComebuyTEA

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

🎄 Build & Learn: Christmas Special- "Agents & the Future of Work"

Join us for Week 5 of the Agentic Flow 7-Week Cycle — Small Extensions & New Tools for Your Agent In Week 5, we keep things simple and supportive. Many people are still shaping their proof-of-concepts, so this session focuses on strengthening what you already built and adding small enhancements that make your workflow more capable.

💡 You can join anytime! Just pick one of the frameworks below and we’ll connect you with others using the same stack so you can learn and build together:

• LangChain / LangGraph (code)
• Crew AI (code)
• OpenAI agent (code)
• n8n (no-code)

Each week we’ll introduce a short lecture or demo on one practical sub-topic (i.e. observability, memory, cost tracking, evaluations etc), then spend the rest of the session working in small groups to experiment and share progress.

Weekly flow:

• 10 min — Intro
• 10 min — Framework-based group recap
• 10 min — Mini-lecture: Extending Your Agent: Tools, Stability & Small Wins [Presentation]
• 90 min— Hands-on build time

Hands On Build Time Choose ONE improvement path: 1️⃣ Add one new tool: API call · file reader · search tool · classifier 2️⃣ Improve stability Guardrails · retries · output validation 3️⃣ Optimize context Shorten instructions · summarize memory · reduce token cost 4️⃣ Reduce latency/cost Caching · fewer model calls · fallback models

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Saturday Dec 13, 10:00 AM- 1 PM 📍 Where? : Joachimsthaler Str. 43, 10623 Berlin ComebuyTEA

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

🎄 Build & Learn: Christmas Special- "Agents & the Future of Work"

Join us for Week 4 of the Agentic Flow 7-Week Cycle — In Week 4, we move from getting things working → making them work better.

You now have a dataset, baseline metric, and a working API connection—so this week we’ll focus on improving a real part of your workflow.

💡 You can join anytime! Just pick one of the frameworks below and we’ll connect you with others using the same stack so you can learn and build together:

• LangChain / LangGraph (code)
• Crew AI (code)
• OpenAI agent (code)
• n8n (no-code)

Each week we’ll introduce a short lecture or demo on one practical sub-topic (i.e. observability, memory, cost tracking, evaluations etc), then spend the rest of the session working in small groups to experiment and share progress.

Weekly flow:

• 10 min — Intro
• 10 min — Framework-based group recap
• 10 min — Mini-lecture: Topic 1: What is MCP?
• 60— Hands-on build time *

Hands-On Build Time

1️⃣ Debug your workflow 2️⃣ Improve one chosen metric (accuracy, speed, cost, consistency)

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Sunday , Nov 30, 10:00 AM- 11:30 AM 📍 Where? : Washingtonpl. 3, 10557 Berlin, Germany

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

🎄 Build & Learn: Christmas Special- "Agents & the Future of Work"

Join us for Week 3 of the Agentic Flow 7-Week Cycle — this week we’ll solidify your project foundations by choosing the right dataset, defining a baseline metric, and getting your authentication + API calls fully working.

💡 You can join anytime! Just pick one of the frameworks below and we’ll connect you with others using the same stack so you can learn and build together:

• LangChain / LangGraph
• Crew AI
• OpenAI API (code)
• Pedantic AI
• n8n (no-code)

Each week we’ll introduce a short lecture or demo on one practical sub-topic (i.e. observability, memory, cost tracking, evaluations etc), then spend the rest of the session working in small groups to experiment and share progress.

Weekly flow:

• 15 min — Intro
• 15 min — Framework-based group recap
• 20 min — Mini-lecture: Topic 1: Choosing datasets, baselines & evaluation metrics, Topic 2: What is MCP?
• 60–90 min — Hands-on build time
• Share & Ask — End with demos + blockers

Week 3 Hands on Building Time 1️⃣ Fix authentication + API access issues 2️⃣ Pick the right dataset or workflow source 3️⃣ Define a baseline + evaluation metric

• What does “good” look like for your workflow?
• Accuracy? Cost? Latency? Success rate?

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Saturday , Nov 22, 10:00 AM- 12:00 PM 📍 Where? : Wework Kemperplatz 1 Mitte D, 10785 Berlin

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

🎄 Build & Learn: Christmas Special- "Agents & the Future of Work"

Join us for Week 2 of the Agentic Flow 7-Week Cycle — we’ll each build a simple working agentic workflow using a framework of our choice and real data.

💡 You can join anytime! Just pick one of the frameworks below and we’ll connect you with others using the same stack so you can learn and build together:

• LangChain / LangGraph
• Crew AI
• OpenAI API (code)
• Pedantic AI
• n8n (no-code)

Each week we’ll introduce a short lecture or demo on one practical sub-topic (i.e. observability, memory, cost tracking, evaluations etc), then spend the rest of the session working in small groups to experiment and share progress.

Weekly flow:

1. 15 min introduction
2. 15 min framework based group recap and share learnings
3. Lecture, demo or topic discussion of the week
4. Group work / implementation

Week 2— Build your first end-to-end working pipeline. We’ll: 1️⃣ Revisit key ideas behind agentic flow (quick recap & Q&A) 2️⃣ Choose or define your dataset/workflow 3️⃣ Build your first end-to-end working pipeline — something small but functional 4️⃣ Share early results and blockers

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Saturday , Nov 8, 10:00 AM- 12:00 PM 📍 Where? : Wework Kemperplatz 1 Mitte D, 10785 Berlin

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

🎄 Build & Learn: Christmas Special- "Agents & the Future of Work" This is our Fifth 7-week cycle. This cycle, instead of everyone building different projects, we’re diving deep together — into one of the most exciting frontiers in AI: Agentic Flow.

🤖 How can we make LLMs reason, plan, and act more like agents? 💰 How do we evaluate cost, performance, or collaboration between agents? 🧩 Which frameworks actually make this easier — LangChain, CrewAI, Autogen, or something new?

Week 1 — Kickoff: What is Agentic Flow?

• Short intro talk + demos of different agent frameworks (LangChain, CrewAI, Autogen, etc.)
• Everyone defines a subtopic or question to explore (e.g., cost tracking, evaluation, multi-agent orchestration).
• Set up environment and pick framework. *

🧭 Possible Individual Topic Ideas

• 📊 Cost Analysis: Track and visualize token usage across frameworks.
• ⚙️ Evaluation: Compare performance of 2 frameworks on the same multi-step task.
• 🧠 Reasoning Styles: Study how different agent setups approach problem-solving.
• 💬 Communication: Design multi-agent conversations for task delegation.
• 💼 Use Case Mini-Build: A practical agent (data summarizer, code reviewer, customer insight bot).

This 7-week themed cycle is about experimenting together — each of us exploring a small slice of the same big question, so we can learn faster by comparing notes.

✨ Who’s Hosting? I’m Lindsey, a senior data scientist working on AI, causal inference, and data products. I’ve built models for fraud detection, uplift modeling, and LLM applications.

📅 When? Saturday , Nov 1, 10:00 AM- 12:00 PM 📍 Where? : Wework Alexanderplatz

💻 Bring: Your laptop, an idea, or just curiosity! 👩‍💻 **No experience needed—just curiosity.**Grab a coffee, meet cool people, and work on something fun.

This week, we’re zooming in on one of the most underrated skills in data science: communicating your project effectively.

Week 4 Focus: Intro to MLOps & Deployment. Topics include packaging your model (move from experimentation to a reproducible, shareable format), setting up basic infrastructure with Docker, FastAPI, or simple cloud deployment, and thinking in pipelines (automating data inputs, retraining, and model monitoring).

Week 3 focus: Intro to Modeling. Now that your data is cleaned and explored, it’s time to build your first model. This session covers training a basic model (regression or classification), evaluating performance with simple metrics, and starting to think about iteration and improvement.

Kickoff and project setup for Week 2 of Build & Learn: Data Science Meetup. Overview of Full Project Track and Self-Learning Track, with guidance on how to start a data science project and prepare for a 7-week journey.

Kickoff for Build & Learn: Data Science Meetup — Week 1. Focus on setting up your tools, getting GitHub ready, choosing a project, and committing to building.

This week’s focus is storytelling and presentation in data science, with emphasis on communicating your project effectively.