Initial commit: MLflow dashboard project

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

routers/experiments.py

@@ -0,0 +1,21 @@
+from typing import List, Optional
+from fastapi import APIRouter, Query
+from schemas import ExperimentSummary, RunSummary
+from services import mlflow_service
+
+router = APIRouter()
+
+
+@router.get("/experiments", response_model=List[ExperimentSummary])
+def list_experiments(
+    tracking_uri: Optional[str] = Query(None, description="MLflow Tracking URI"),
+):
+    return mlflow_service.get_experiments(tracking_uri)
+
+
+@router.get("/experiments/{exp_id}/runs", response_model=List[RunSummary])
+def list_runs(
+    exp_id: str,
+    tracking_uri: Optional[str] = Query(None, description="MLflow Tracking URI"),
+):
+    return mlflow_service.get_runs(tracking_uri, exp_id)

routers/runs.py

@@ -0,0 +1,23 @@
+from fastapi import APIRouter, Query
+from typing import Optional
+from schemas import RunDetail, MLflowLink
+from services import mlflow_service
+
+router = APIRouter()
+
+
+@router.get("/runs/{run_id}", response_model=RunDetail)
+def get_run(
+    run_id: str,
+    tracking_uri: Optional[str] = Query(None, description="MLflow Tracking URI"),
+):
+    return mlflow_service.get_run_detail(tracking_uri, run_id)
+
+
+@router.get("/runs/{run_id}/mlflow-link", response_model=MLflowLink)
+def get_mlflow_link(
+    run_id: str,
+    tracking_uri: Optional[str] = Query(None, description="MLflow Tracking URI"),
+):
+    url = mlflow_service.get_mlflow_link(tracking_uri, run_id)
+    return {"url": url}

routers/serve.py

@@ -0,0 +1,73 @@
+import subprocess
+import uuid
+from typing import Dict, List
+from fastapi import APIRouter, HTTPException
+from schemas import ServeRequest, ServeStatus
+
+router = APIRouter()
+
+# In-memory store for serving processes
+_serving_processes = {}  # type: Dict[str, Dict]
+
+
+@router.post("/serve", response_model=ServeStatus)
+def start_serve(req: ServeRequest):
+    serve_id = uuid.uuid4().hex[:8]
+    cmd = [
+        "mlflow", "models", "serve",
+        "-m", req.model_uri,
+        "-p", str(req.port),
+        "--no-conda",
+    ]
+    if req.tracking_uri:
+        cmd.extend(["--env-manager", "local"])
+
+    try:
+        proc = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    except FileNotFoundError:
+        raise HTTPException(status_code=500, detail="mlflow CLI not found")
+
+    _serving_processes[serve_id] = {
+        "id": serve_id,
+        "model_uri": req.model_uri,
+        "port": req.port,
+        "pid": proc.pid,
+        "process": proc,
+    }
+
+    return ServeStatus(
+        id=serve_id,
+        model_uri=req.model_uri,
+        port=req.port,
+        pid=proc.pid,
+        status="running",
+    )
+
+
+@router.get("/serve", response_model=List[ServeStatus])
+def list_serve():
+    results = []
+    for sid, info in _serving_processes.items():
+        proc = info["process"]
+        status = "running" if proc.poll() is None else "stopped"
+        results.append(ServeStatus(
+            id=sid,
+            model_uri=info["model_uri"],
+            port=info["port"],
+            pid=info["pid"],
+            status=status,
+        ))
+    return results
+
+
+@router.delete("/serve/{serve_id}")
+def stop_serve(serve_id: str):
+    if serve_id not in _serving_processes:
+        raise HTTPException(status_code=404, detail="Serve process not found")
+
+    proc = _serving_processes[serve_id]["process"]
+    if proc.poll() is None:
+        proc.terminate()
+
+    model_uri = _serving_processes.pop(serve_id)["model_uri"]
+    return {"message": f"Stopped serving {model_uri}"}

routers/train.py

@@ -0,0 +1,11 @@
+from fastapi import APIRouter, HTTPException
+
+router = APIRouter()
+
+
+@router.post("/train")
+def start_train():
+    raise HTTPException(
+        status_code=501,
+        detail="Train trigger is not implemented yet. Will be connected to training server.",
+    )