Week 4 Langgraph Career Sprint Submission

4_langgraph/community_contributions/sammyloto/.env.example

@@ -0,0 +1,14 @@
+# Required
+OPENAI_API_KEY=
+
+# Optional: defaults to gpt-4o-mini
+OPENAI_MODEL=gpt-4o-mini
+
+# Optional: better web search than the free fallback (DuckDuckGo)
+# Get a key at https://serper.dev
+SERPER_API_KEY=
+
+# LangSmith (optional, Week 4 observability)
+# LANGCHAIN_TRACING_V2=true
+# LANGCHAIN_API_KEY=
+# LANGCHAIN_PROJECT=sammyloto-career-sprint

4_langgraph/community_contributions/sammyloto/.gitignore

@@ -0,0 +1,7 @@
+__pycache__/
+*.py[cod]
+.env
+checkpoints/
+*.db
+sandbox/*
+!sandbox/.gitkeep

4_langgraph/community_contributions/sammyloto/README.md

@@ -0,0 +1,60 @@
+# Career Sprint Sidekick (Week 4 community contribution)
+
+A **bounded career task** assistant: research a company and role, align with a local profile, and produce talking points plus a short outreach draft - validated by a **worker -> tools -> evaluator** loop and **SQLite checkpointing**. Built for *Master AI Agents in 30 days* (LangGraph week).
+
+## What this project demonstrates
+
+
+| Week 4 theme                    | Here                                                                                                           |
+| ------------------------------- | -------------------------------------------------------------------------------------------------------------- |
+| **Days 1-2 - Graph + state**    | `CareerSprintState` (`TypedDict`) with `add_messages` reducer; explicit nodes and edges.                       |
+| **Day 3 - Persistence**         | `AsyncSqliteSaver` -> `checkpoints/career_sprint.db` (WAL mode).                                               |
+| **Day 3 - Conditional routing** | Worker routes to `ToolNode` or evaluator; evaluator routes to `END` or back to worker.                         |
+| **Day 4 - Structured outputs**  | `SprintRubricAssessment` (Pydantic) for the evaluator; Gradio UI.                                              |
+| **Day 5 - Tools**               | Web search (Serper or DuckDuckGo fallback), `read_career_profile`, Wikipedia, Python REPL, sandbox file tools. |
+
+
+**Not included:** Playwright browser automation (optional in the course). Search + file + profile read cover the "tooling" learning goals without a browser install.
+
+## Layout
+
+- `graph.py` - LangGraph definition, default sprint rubric, `CareerSprintGraph.run_turn`.
+- `tools.py` - Tool wiring; sandbox root is `sandbox/`; optional profile is `me/summary.txt`.
+- `app.py` - Gradio app: **one `thread_id` per UI session** (reset starts a new thread and new compiled app instance).
+- `checkpoints/` - Created at runtime (gitignored) for SQLite checkpoints.
+
+## Setup
+
+```bash
+cd 4_langgraph/community_contributions/sammyloto_career_sprint
+python3 -m venv .venv && source .venv/bin/activate   # optional
+pip install -r requirements.txt
+cp .env.example .env
+# Set OPENAI_API_KEY; optionally SERPER_API_KEY for stronger search
+```
+
+## Run
+
+```bash
+python app.py
+```
+
+Open the URL shown (default `http://127.0.0.1:7860`). Fill **company** and **role**, adjust **success criteria** if you like, then send a sprint request.
+
+## Optional: personal profile (continuity with Week 1 style RAG)
+
+Add a short bio under `me/summary.txt` (skills, goals, voice). The `read_career_profile` tool lets the worker tailor pitches without bundling Chroma in this submission.
+
+## Optional: LangSmith
+
+To trace runs in LangSmith, set in `.env`:
+
+```env
+LANGCHAIN_TRACING_V2=true
+LANGCHAIN_API_KEY=your_key
+LANGCHAIN_PROJECT=sammyloto-career-sprint
+```
+
+## Attribution
+
+Patterns are **inspired by** the course Sidekick lab (`4_lab4` / `sidekick.py`)

4_langgraph/community_contributions/sammyloto/app.py

@@ -0,0 +1,165 @@
+"""Gradio UI: isolated thread IDs + SQLite-backed LangGraph runs."""
+
+from __future__ import annotations
+
+import os
+import sys
+from typing import Optional
+
+import gradio as gr
+from dotenv import load_dotenv
+from langchain_core.messages import AIMessage, BaseMessage, HumanMessage, ToolMessage
+
+# Ensure this folder is on path when executed from repo root
+_ROOT = os.path.dirname(os.path.abspath(__file__))
+if _ROOT not in sys.path:
+    sys.path.insert(0, _ROOT)
+
+from graph import CareerSprintGraph, DEFAULT_SPRINT_CRITERIA, new_thread_id
+
+load_dotenv(override=True)
+
+
+def messages_to_chat_ui(messages: list[BaseMessage]) -> list[dict]:
+    """Flatten LangChain messages into Gradio chatbot dicts."""
+    out: list[dict] = []
+    for m in messages:
+        if isinstance(m, HumanMessage):
+            out.append({"role": "user", "content": m.content or ""})
+        elif isinstance(m, AIMessage):
+            text = (m.content or "").strip() or "[model used tools]"
+            out.append({"role": "assistant", "content": text})
+        elif isinstance(m, ToolMessage):
+            snippet = (m.content or "")[:1200]
+            out.append({"role": "assistant", "content": f"_(tool result)_\n{snippet}"})
+    return out
+
+
+async def setup_app() -> tuple[CareerSprintGraph, str, list]:
+    app = CareerSprintGraph()
+    await app.setup()
+    tid = new_thread_id()
+    return app, tid, []
+
+
+async def on_message(
+    app: CareerSprintGraph,
+    thread_id: str,
+    lc_history: list[BaseMessage],
+    user_text: str,
+    success_criteria: str,
+    company: str,
+    role: str,
+):
+    if app is None:
+        err = [
+            {
+                "role": "assistant",
+                "content": "App is still loading. Wait until startup finishes, then try again.",
+            }
+        ]
+        return err, lc_history, thread_id, app
+    if not user_text or not str(user_text).strip():
+        return gr.update(), lc_history, thread_id, app
+
+    result = await app.run_turn(
+        user_text.strip(),
+        success_criteria,
+        company,
+        role,
+        lc_history,
+        thread_id,
+    )
+    new_lc: list[BaseMessage] = list(result["messages"])
+    chat = messages_to_chat_ui(new_lc)
+    return chat, new_lc, thread_id, app
+
+
+async def on_reset(_app: Optional[CareerSprintGraph]):
+    """New graph instance + thread so prior checkpoints do not leak into the next demo."""
+    new_app = CareerSprintGraph()
+    await new_app.setup()
+    tid = new_thread_id()
+    if _app is not None:
+        await _app.aclose()
+    return (
+        new_app,
+        tid,
+        [],
+        [],
+        gr.update(value=""),
+        gr.update(value=""),
+        gr.update(value=""),
+        gr.update(value=DEFAULT_SPRINT_CRITERIA),
+    )
+
+
+def main() -> None:
+    with gr.Blocks(
+        title="Career Sprint Sidekick",
+        theme=gr.themes.Default(primary_hue="teal"),
+    ) as demo:
+        gr.Markdown(
+            "## Career Sprint Sidekick\n"
+            "LangGraph + tools + rubric loop + SQLite checkpoints. "
+            "Optional **me/summary.txt** personalizes output. Writes packs under **sandbox/**."
+        )
+        app_state = gr.State()
+        thread_state = gr.State()
+        lc_state = gr.State([])
+
+        with gr.Row():
+            company = gr.Textbox(label="Target company", placeholder="e.g. Acme Labs")
+            role = gr.Textbox(label="Target role", placeholder="e.g. Backend Engineer")
+        success_criteria = gr.Textbox(
+            label="Success criteria (optional)",
+            lines=4,
+            value=DEFAULT_SPRINT_CRITERIA,
+        )
+        chat = gr.Chatbot(label="Sprint thread", height=420)
+        user_in = gr.Textbox(
+            label="Your message",
+            placeholder="Describe the sprint, e.g. research + talking points + outreach note...",
+            lines=2,
+        )
+        with gr.Row():
+            go = gr.Button("Run sprint turn", variant="primary")
+            reset = gr.Button("New session", variant="secondary")
+
+        demo.load(
+            setup_app,
+            [],
+            [app_state, thread_state, lc_state],
+            time_limit=180,
+        )
+
+        go.click(
+            on_message,
+            [app_state, thread_state, lc_state, user_in, success_criteria, company, role],
+            [chat, lc_state, thread_state, app_state],
+            time_limit=600,
+            show_progress="minimal",
+        )
+        user_in.submit(
+            on_message,
+            [app_state, thread_state, lc_state, user_in, success_criteria, company, role],
+            [chat, lc_state, thread_state, app_state],
+            time_limit=600,
+            show_progress="minimal",
+        )
+        reset.click(
+            on_reset,
+            [app_state],
+            [app_state, thread_state, lc_state, chat, user_in, company, role, success_criteria],
+            time_limit=180,
+        )
+
+    demo.queue()
+
+    host = os.getenv("GRADIO_SERVER_NAME", "127.0.0.1")
+    port = int(os.getenv("GRADIO_SERVER_PORT", "7860"))
+    demo.launch(server_name=host, server_port=port, share=False)
+
+
+if __name__ == "__main__":
+    main()