مین مواد پر جائیں

Chapter 40: Class-Based aur Home-Based Projects

MODULE 6 | MLOps​

Chapter 40: Class-Based aur Home-Based Projects​

Ab tak har chapter me ek tool seekha: FastAPI , Docker , MLflow , CI/CD , Logging . Is chapter me ye sab ek asli project me jurte hain: AI Medical Assistant . User apni symptoms chunta hai , aur app top 3 mumkina bimariyan aur unka imkaan dikhati hai. Isme FastAPI backend , khoobsurat web UI , tests , Docker aur CI sab hain.

ETHICS / ZIMMEDARI​

Ye project sirf seekhne ke liye hai. Ye doctor ya tashkhees (diagnosis) ka badal nahi . Hum ne banawati (synthetic) data use kiya hai, jo asli marizon ka nahi. Asli medical AI ke liye sahi dataset, doctors ki rehnumai, bohat sakht testing aur qanooni manzoori chahiye. App ke har jawab ke sath disclaimer dikhana is project ka hissa hai, koi aakhri khoobi nahi. Aur kabhi kisi ko bimari ki tashkhees ya dawa ka mashwara is app ke bharose par mat do.

40.1 Project ka maqsad aur dayra​

PROBLEM STATEMENT​

Bohat se log chhoti bimari par pehle internet par symptoms dhoondte hain. Hum ek seekhne wala symptom checker banate hain: user symptoms chunta hai, aur API top 3 mumkina conditions + imkaan (probability) wapas karti hai. Saath hi saaf likha hota hai ke ye doctor ka mashwara nahi.

Input: symptoms ki list (maslan fever, headache). Output: top 3 conditions + disclaimer. Model: Random Forest. Backend: FastAPI. UI: ek HTML page.

DIAGRAM​
Browser (UI) ----> FastAPI ----> model.joblib
static/index.html /predict (Random Forest)
| |
| logging (har request)
v v
Docker container <- CI (GitHub Actions: test + build)

Project ka folder aise bana hoga:

FOLDER KA NAQSHA medical-assistant/ train.py <- data banao, model train aur save karo train_mlflow.py <- (Home Project A) MLflow ke sath training requirements.txt <- API ki libraries (versions ke sath) requirements-dev.txt <- testing ki libraries Dockerfile .dockerignore .gitignore app/ __init__.py <- khali file

main.py <- FastAPI backend
static/
index.html <- UI
model/ <- train.py yahan model.joblib aur symptoms.json rakhta hai
tests/
test_api.py
.github/workflows/
ci.yml <- (Chapter 38 wali file)

40.2 Class-Based Project: Qadam 1, Data aur Model​

Asli medical dataset hamare paas nahi (aur hona bhi chahiye to ijazat chahiye), isliye train.py me 9 aam bimariyon aur 18 symptoms ka ek banawati dataset banta hai: har bimari ke liye likha hai ke har symptom hone ka kitna imkaan hai (maslan Flu me fever ka 0.90). Phir har marz ke 300 sample random banate hain aur Random Forest train karte hain.

train.py

import json
import os
import joblib
import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
SYMPTOMS = [ "fever", "cough", "sore_throat", "runny_nose", "sneezing", "headache", "body_ache", "fatigue", "nausea", "vomiting", "diarrhea", "stomach_pain", "rash", "chills", "joint_pain", "light_sensitivity", "itchy_eyes", "loss_of_appetite", ] # Har bimari ke liye:
symptom -> us ke hone ka imkaan (0 se 1)
PROFILES = { "Common Cold": {"runny_nose": .9, "sneezing": .8, "sore_throat": .6, "cough": .5, "headache": .3, "fatigue": .3, "fever": .15}, "Flu": {"fever": .9, "body_ache": .85, "chills": .7, "cough": .7, "fatigue": .8, "headache": .6, "sore_throat": .4}, "Dengue": {"fever": .95, "headache": .8, "joint_pain": .8, "body_ache": .8, "rash": .5, "nausea": .5, "fatigue": .7, "chills": .4, "loss_of_appetite": .5}, "Typhoid": {"fever": .9, "stomach_pain": .7, "headache": .6, "loss_of_appetite": .7, "fatigue": .7, "nausea": .4, "diarrhea": .4}, "Malaria": {"fever": .9, "chills": .9, "headache": .6, "nausea": .5, "vomiting": .4, "body_ache": .5, "fatigue": .6}, "Migraine": {"headache": .98, "light_sensitivity": .85, "nausea": .6, "vomiting": .3, "fatigue": .3}, "Food Poisoning": {"nausea": .85, "vomiting": .85, "diarrhea": .85, "stomach_pain": .8, "fever": .3, "fatigue": .4}, "Allergy": {"sneezing": .85, "itchy_eyes": .85, "runny_nose": .7, "rash": .3, "cough": .3}, "Gastritis": {"stomach_pain": .9, "nausea": .6, "vomiting": .3, "loss_of_appetite": .6}, }
def make_dataset(per_class=300, seed=42):
rng = np.random.default_rng(seed)
rows, labels = [], []
for disease, profile in PROFILES.items():
for _ in range(per_class):
row = [int(rng.random() < profile.get(s, 0.03)) for s in SYMPTOMS]
rows.append(row)
labels.append(disease)
return pd.DataFrame(rows, columns=SYMPTOMS), pd.Series(labels, name="disease")
if __name__ == "__main__":
X, y = make_dataset()
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, stratify=y, random_state=42)
model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train, y_train)
acc = accuracy_score(y_test, model.predict(X_test))
print(f"Rows: {len(X)} | Symptoms: {len(SYMPTOMS)} | Diseases: {y.nunique()}")
print(f"Test accuracy: {acc:.2%}")
os.makedirs("model", exist_ok=True)
joblib.dump(model, "model/model.joblib")
with open("model/symptoms.json", "w") as f:
json.dump(SYMPTOMS, f)
print("Saved: model/model.joblib, model/symptoms.json")
TERMINAL​
python train.py
OUTPUT​
Rows: 2700 | Symptoms: 18 | Diseases: 9
Test accuracy: 88.89%
Saved: model/model.joblib, model/symptoms.json
HissaKya karta hai
SYMPTOMS18 symptoms kitarteeb wali list; model ke columns isi tarteeb me hain
PROFILESHar bimari ke liye symptom kaimkaan; jo symptom list me nahi, uska imkaan0.03(kabhi
kabhar aam wajah se)
make_dataset()default_rng(42) sehamesha ek jaisadata banata hai (seed), taake nateeja dobara mil sake
stratify=yTrain aur test dono mehar bimari ka barabar hissa
joblib.dumpModel file model/model.joblib me, aur symptoms ki tarteeb symptoms.json me
AAM GALTIYAN​
  • 88.89% accuracy ka matlab ye nahi ke ye asli duniya me kaam karegi. Data hamne apni marzi ke imkaan se banaya, to model sirf wohi imkaan seekh raha hai .

  • Aage dekhoge ke model aksar 100% (1.0) confidence deta hai. Ye bhi banawati, aasan data ki nishani hai. Asli medical data me aise nahi hota.

40.3 Qadam 2: FastAPI Backend​

Ab app/main.py me wo sab jo Chapter 35 aur 39 me seekha: model ek baar load , Pydantic se input check, logging , aur har jawab ke sath disclaimer :

app/main.py

import json
import logging
import time
from pathlib import Path
import joblib
import pandas as pd
from fastapi import FastAPI, HTTPException
from fastapi.responses import FileResponse
from pydantic import BaseModel, Field
BASE_DIR = Path(__file__).resolve().parent.parent logging.basicConfig( level=logging.INFO, format="%(asctime)s | %(levelname)s | %(name)s | %(message)s", )
logger = logging.getLogger("medical-assistant")
model = joblib.load(BASE_DIR / "model" / "model.joblib")
SYMPTOMS = json.loads((BASE_DIR / "model" / "symptoms.json").read_text())
DISCLAIMER = "Yeh sirf seekhne ke liye hai. Yeh doctor ka mashwara nahi hai."
app = FastAPI(title="AI Medical Assistant", version="1.0.0")
class PredictRequest(BaseModel):
symptoms:
list[str] = Field(min_length=1, description="Symptoms ke naam")
class Prediction(BaseModel):
disease:
str probability:
float
class PredictResponse(BaseModel):
predictions:
list[Prediction] disclaimer:
str @app.get("/")
def home():
return FileResponse(BASE_DIR / "static" / "index.html") @app.get("/health")
def health():
return {"status": "ok"} @app.get("/symptoms")
def list_symptoms():
return {"symptoms": SYMPTOMS}
@app.post("/predict", response_model=PredictResponse)
def predict(req: PredictRequest):
start = time.perf_counter()
unknown = [s for s in req.symptoms if s not in SYMPTOMS]
if unknown:
logger.warning("Unknown symptoms: %s", unknown)
raise HTTPException(status_code=400,
detail=f"Ye symptoms nahi milay: {unknown}")
values = [1 if s in req.symptoms else 0 for s in SYMPTOMS]
row = pd.DataFrame([values], columns=SYMPTOMS)
probs = model.predict_proba(row)[0]
top3 = sorted(zip(model.classes_, probs), key=lambda p: p[1], reverse=True)[:3]
ms = (time.perf_counter() - start) * 1000
logger.info("predict | symptoms=%d | top=%s | %.1f ms",
len(req.symptoms), top3[0][0], ms)
return PredictResponse(
predictions=[Prediction(disease=d, probability=round(float(p), 4))
for d, p in top3],
disclaimer=DISCLAIMER,
)
EndpointMethodKaam
/GETUI ka page(static/index.html) dikhata hai
/healthGETZindahone ka check (Docker, CI aur monitoring ke liye)
/symptomsGETModel jin18 symptomsko jaanta hai unki list (UI isi se chips banata hai)
/predictPOSTSymptoms bhejo,top 3 conditions + probability + disclaimerpao
Code ka khaas hissaKyun hai
min_length=1 (PredictRequest)Khali listpar khud422(kam az kam ek symptom zaroor)
unknown = [...] aur HTTPException(400)Jo symptom model ko aata hi nahi, us parsaaf 400jawab aurWARNING
log
pd.DataFrame([values], columns=SYMPTOMS)Model kocolumn ke naamonke sath do (training me bhi naam the)
predict_proba + sorted(...)[:3]Har bimari kaimkaanle kar sab se bare3chunna
logger.info(...)Har prediction kiginti, top nateeja aur waqt(nijji symptoms nahi)
DISCLAIMERHar jawab mewarning, taake UI ya koi bhi client ise na chhor sake

40.4 Qadam 3: Web UI​

Ab sab se khoobsurat hissa: ek web page jahan user symptoms chips ki tarah dabata hai aur Check karta hai. Ye ek akeli HTML file hai (static/index.html), koi framework nahi. Rang wohi hain jo is book ke hain ( pink aur lavender ):

static/index.html <!DOCTYPE html> <html lang="en"> <head> <meta charset="UTF-8"> <meta name="viewport" content="width=device-width, initial-scale=1"> <title>AI Medical Assistant</title> <style> body { font-family: Arial, sans-serif; background: #FDEBF2; color: #2b2b33; margin: 0; }

.wrap { max-width: 640px; margin: 0 auto; padding: 24px 16px; }
h1 { color: #A4457A; margin-bottom: 4px; }
.card { background: #fff; border: 1px solid #F4C6D7; border-radius: 12px;
padding: 16px; margin-top: 16px; }
.chip { display: inline-block; padding: 6px 12px; margin: 4px;
border: 1px solid #F4C6D7;
border-radius: 20px; cursor: pointer; background: #fff; }
.chip.on { background: #A4457A; color: #fff; border-color: #A4457A; }
button { background: #A4457A; color: #fff; border: 0; padding: 10px 20px;
border-radius: 8px;
font-size: 16px; cursor: pointer; margin-top: 12px; }
.bar { background: #F4C6D7; border-radius: 6px; height: 10px; margin-top: 4px; }
.fill { background: #A4457A; height: 10px; border-radius: 6px; }
.warn { font-size: 13px; color: #6b6b75; margin-top: 12px; }
</style>
</head>
<body>
<div class="wrap">
<h1>AI Medical Assistant</h1>
<div>Apni symptoms chunein, phir Check karein.</div>
<div class="card">
<div id="chips">Loading...</div>
<button onclick="check()">Check</button>
</div>
<div class="card" id="result" style="display:none"></div>
<div class="warn" id="disclaimer"></div>
</div>
<script>
const selected = new Set();
async function loadSymptoms() {
const res = await fetch("/symptoms");
const data = await res.json();
const box = document.getElementById("chips");
box.innerHTML = "";
data.symptoms.forEach(s => {
const el = document.createElement("span");
el.className = "chip";
el.textContent = s.replaceAll("_", " ");
el.onclick = () => {
selected.has(s) ? selected.delete(s) : selected.add(s);
el.classList.toggle("on");
};
box.appendChild(el);
});
}
async function check() {
const result = document.getElementById("result");
if (selected.size === 0) {
result.style.display = "block";
result.textContent = "Pehle kam az kam ek symptom chunein.";
return;
}
const res = await fetch("/predict", {
method: "POST",
headers: { "Content-Type": "application/json" },
body: JSON.stringify({ symptoms: Array.from(selected) })
});
const data = await res.json();
if (!res.ok) {
result.style.display = "block";
result.textContent = data.detail;
return;
}
result.style.display = "block";
result.innerHTML = "<b>Top 3 imkaanat</b>" + data.predictions.map(p =>
`<div style="margin-top:10px">${p.disease}: ${(p.probability * 100).toFixed(1)}%
<div class="bar">
<div class="fill" style="width:${p.probability * 100}%"></div>
</div></div>`
).join("");
document.getElementById("disclaimer").textContent = data.disclaimer;
}
loadSymptoms();
</script>
</body>
</html>
UI ka hissaKya karta hai
loadSymptoms()Page khulte hi**/symptomsse list mangta hai aur har symptom ka ekchip**banata hai
selected (Set)User nekaun se chipsdabaye, unka record
el.onclickChip dabane parchuna / hata; on class se rang badalta hai
check()POST /predictkarta hai, aur jawab mebarske sath top 3 dikhata hai
Error handlingKuch na chuna ho ya server400de tosaaf messagedikhata hai
disclaimerJawab ke neechewarninghamesha dikhti hai
PRACTICAL TIP​

UI ka fetch("/predict") bina poora address likhe kaam karta hai kyunke UI aur API ek hi server se aate hain (/ route UI deta hai). Isliye CORS ka masla nahi aata. Agar UI ko kisi alag jagah (maslan Vercel) par rakho, to API me CORS lagana parega.

40.5 Qadam 4: Chalana aur Test karna​

TERMINAL​
python train.py
uvicorn app.main:app --reload

Browser me http://127.0.0.1:8000 kholo: UI nazar aayegi. /docs par Swagger testing page hai. curl se bhi dekho (symptoms: fever, headache, joint_pain, body_ache):

TERMINAL​
curl -X POST http://127.0.0.1:8000/predict -H "Content-Type: application/json" \
-d '{"symptoms":["fever","headache","joint_pain","body_ache"]}'

OUTPUT (asal jawab ek hi line me aata hai; parhne ke liye todi hai) {"predictions":[{"disease":"Dengue","probability":0.9967}, {"disease":"Malaria","probability":0.0033}, {"disease":"Allergy","probability":0.0}], "disclaimer":"Yeh sirf seekhne ke liye hai. Yeh doctor ka mashwara nahi hai."}

Galat input par bhi dekho ke app sahi tarah rok deti hai:

OUTPUT (galat symptom aur khali list)​
{"detail":"Ye symptoms nahi milay: ['xyz']"} <- status 400
{"detail":[{"type":"too_short","loc":["body","symptoms"], ... }]} <- status 422 (khali
list)

Ab tests (tests/test_api.py) likhte hain. Ye Chapter 35 aur 38 me seekhe hue TestClient + pytest hain:

tests/test_api.py

from fastapi.testclient import TestClient
from app.main import app
client = TestClient(app)
def test_health():
res = client.get("/health")
assert res.status_code == 200
assert res.json() == {"status": "ok"}
def test_symptoms_list():
res = client.get("/symptoms")
assert res.status_code == 200
assert "fever" in res.json()["symptoms"]
def test_predict_returns_top3():
res = client.post("/predict", json={"symptoms": ["nausea", "vomiting", "diarrhea"]})
assert res.status_code == 200
preds = res.json()["predictions"]
assert len(preds) == 3
assert preds[0]["disease"] == "Food Poisoning"
def test_predict_unknown_symptom():
res = client.post("/predict", json={"symptoms": ["xyz"]})
assert res.status_code == 400
def test_predict_empty_list():
res = client.post("/predict", json={"symptoms": []})
assert res.status_code == 422
TERMINAL​
pytest -v

OUTPUT

tests/test_api.py::test_health PASSED [ 20%]
tests/test_api.py::test_symptoms_list PASSED [ 40%]
tests/test_api.py::test_predict_returns_top3 PASSED [ 60%]
tests/test_api.py::test_predict_unknown_symptom PASSED [ 80%]
tests/test_api.py::test_predict_empty_list PASSED [100%]
========================= 5 passed, 1 warning in 1.58s =========================

40.6 Qadam 5: Docker aur CI​

Ab project ko dabbe me band karte hain (Chapter 36). Pehle versions ke sath requirements.txt (apne pip freeze ke versions likho; ye wo hain jin par book ka code test hua):

requirements.txt

fastapi==0.159.1

uvicorn==0.54.0 scikit-learn==1.8.0 joblib==1.5.3 pandas==3.0.2 numpy==2.4.4 requirements-dev.txt pytest==9.1.1 httpx==0.28.1 Dockerfile FROM python:3.12-slim WORKDIR /app COPY requirements.txt . RUN pip install --no-cache-dir -r requirements.txt COPY app/ app/ COPY static/ static/ COPY model/ model/ EXPOSE 8000

CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8000"] TERMINAL docker build -t medical-assistant . docker run -d -p 8000:8000 --name medical-container medical-assistant docker ps MISAAL OUTPUT (ID aur waqt alag honge) CONTAINER ID IMAGE STATUS PORTS NAMES b7c3d9e1f2a4 medical-assistant Up 8 seconds 0.0.0.0:8000->8000/tcp medical-container

http://localhost:8000 par wohi UI container ke andar se chalegi. Chhota model ( takreeban 9 MB ) hone ki wajah se wo seedha image me COPY model/ model/ se chala jata hai aur GitHub par bhi push ho jata hai.

Phir CI (Chapter 38): .github/workflows/ci.yml wohi file jo Chapter 38 me di gayi. Project GitHub par push karo, Actions tab me test aur docker-build dono jobs ka hara tick dekho, aur README me CI badge lagao.

AAM GALTIYAN​
  • Python ya scikit-learn ka version alag hone par model load nahi hota ya galat nateeja deta hai. Docker aur CI me wohi version pin karo jo training me tha.

  • model/ folder ko .gitignore me daal dena : phir CI aur Docker me model nahi milta aur build toot jata hai.

  • app/__init__.py file banana bhool jana (

from app.main ...

ka import masla).

  • UI test karte waqt server ka chalte hona zaroori hai; index.html ko seedha double-click kar ke kholne par fetch nahi chalta.

40.7 Class-Based Project: Deliverables aur Rubric​

Maqsad: Class me upar ke saare qadam khud chalana aur ek kaam karta hua AI Medical Assistant jama karana.

Deliverables​

  • Chalne wala project ( GitHub repo ) jis me train.py, app/, static/, tests/, Dockerfile, requirements.txt aur ci.yml hon.

  • README.md : project kya karta hai, kaise chalaen (3 se 4 command), screenshots (UI aur /docs), limitations aur disclaimer .

  • Screenshots : UI par result, Swagger /docs, pytest ka "5 passed", aur GitHub Actions ka hara tick .

  • Docker se chalta hua container ka screenshot (docker ps aur UI).

  • Demo: 2 se 3 minute ki screen recording.

RubricNumber
Model train hota hai aur API sahi prediction deti hai (/predict, /health, /symptoms)25
UI chalti hai (chips, Check, top 3 bars, error message)20
Tests likhe hain aur pass hote hain (kam az kam 5, ek galat input ka)15
Docker image build hoti aur container chalta hai15
CI (GitHub Actions) hara hai, README aur screenshots saaf hain15
Ethics: disclaimer,nijji data logs me nahi, limitations imandari se likhi hain10
Kul100

40.8 Home-Based Project A: MLflow tracking aur Model Registry​

Maqsad: training ka hisaab MLflow me rakho (Chapter 37) aur sab se behtar model ko registry me champion banao. train_mlflow.py train.py se make_dataset() import karta hai aur teen settings chalata hai:

train_mlflow.py​
import mlflow
import mlflow.sklearn
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score
from sklearn.model_selection import train_test_split
from train import make_dataset
mlflow.set_tracking_uri("sqlite:///mlflow.db")
mlflow.set_experiment("medical-assistant")
X, y = make_dataset()
X_train, X_test, y_train, y_test = train_test_split(
X, y, test_size=0.2, stratify=y, random_state=42)
for n_trees, depth in [(25, 8), (50, 12), (100, None)]:
with mlflow.start_run(run_name=f"rf_{n_trees}_{depth}"):
model = RandomForestClassifier(n_estimators=n_trees, max_depth=depth,
random_state=42)
model.fit(X_train, y_train)
acc = accuracy_score(y_test, model.predict(X_test))
mlflow.log_params({"n_estimators": n_trees, "max_depth": depth})
mlflow.log_metric("accuracy", acc)
mlflow.sklearn.log_model(model, name="model",
serialization_format="cloudpickle")
print(f"n_estimators={n_trees:>3}, max_depth={str(depth):>4} "
f"-> accuracy={acc:.4f}")
TERMINAL​

python train_mlflow.py mlflow ui --backend-store-uri sqlite:///mlflow.db OUTPUT (pehli command ka; MLflow ki INFO lines chhor di hain)

n_estimators= 25, max_depth= 8 ->
accuracy=0.8815
n_estimators= 50, max_depth= 12 ->
accuracy=0.8852
n_estimators=100, max_depth=None ->
accuracy=0.8889

Karne ke kaam:​

  1. UI (http://127.0.0.1:5000) me teeno runs compare karo. Kya 100 trees ka faida waqai kafi hai ya 25 trees (chhota, tez model) kaafi tha?

  2. train_accuracy bhi log karo aur dekho train aur test accuracy me kitna farq hai (overfitting).

  3. Har run ke sath model ki file size (MB) bhi metric ke tor par log karo aur size aur accuracy ka sauda (tradeoff) likho.

  4. Sab se behtar run ko register_model se medical-assistant-model naam se register karo aur champion alias lagao.

  5. Mushkil: app/main.py me model models:/medical-assistant-model@champion se load karwao (MLflow install hona zaroori hoga) aur sochkar likho ke Docker image me ye kaise chalega.

NOTE

mlflow.db aur mlruns/ ko .gitignore aur .dockerignore me rakho; ye bare ho sakte hain aur API ko in ki zaroorat nahi.

40.9 Home-Based Project B: Skin Image ka /predict/image endpoint​

Maqsad: App ko do tarah ke input dena: symptoms likhna ya jildi masle ki tasveer upload karna. Iske liye tumhe pehle apna image model train karna hoga (Module 5: transfer learning, maslan MobileNetV2 par DermNet jese dataset). Training ke baad do cheezein save karo: model/skin_model.h5 aur model/class_indices.json (Keras me train_generator.class_indices, jo naam -> index deta hai). Phir app/image_model.py banao:

app/image_model.py

import io
import json
from pathlib import Path
import numpy as np
from PIL import Image
BASE_DIR = Path(__file__).resolve().parent.parent
_model = None
_classes = None
def _load():
global _model, _classes
if _model is None:
from tensorflow.keras.models import load_model # sirf zaroorat par
import
_model = load_model(BASE_DIR / "model" / "skin_model.h5")
mapping = json.loads((BASE_DIR / "model" / "class_indices.json").read_text())
_classes = {index: name for name, index in mapping.items()}
return _model, _classes
def predict_skin(image_bytes: bytes, top_k: int = 3):
model, classes = _load()
img = Image.open(io.BytesIO(image_bytes)).convert("RGB").resize((224, 224))
arr = np.asarray(img, dtype="float32") / 255.0
probs = model.predict(arr[None, ...], verbose=0)[0]
top = probs.argsort()[::-1][:top_k]
return [{"condition": classes[int(i)], "probability": round(float(probs[i]), 4)} for i in top]

Phir app/main.py me ye endpoint jodo (upar

from fastapi import File, UploadFile

aur

from app.image_model import predict_skin

bhi likho):

app/main.py (naya endpoint)​
@app.post("/predict/image")
async def predict_image(file: UploadFile = File(...)):
if file.content_type not in ("image/jpeg", "image/png"):
raise HTTPException(status_code=400, detail="Sirf JPG ya PNG tasveer bhejo")
data = await file.read()
logger.info("predict/image | %d bytes", len(data))
return {"predictions": predict_skin(data), "disclaimer": DISCLAIMER}

Pehle se zaroori cheezein: requirements.txt me tensorflow, Pillow aur python-multipart jodo (file upload ke liye python-multipart chahiye). Image ko wohi preprocessing do jo training me thi (yahan 224x224, RGB, /255.0); agar training me preprocess_input use hua tha to yahan bhi wohi lagao, warna nateeje kharab aayenge.

VERSION NOTE​

Is endpoint ka logic (type check, resize, top 3, jawab ka format) ek nakli model ke sath test hua hai: tasveer par 200 , aur text file par 400 . Lekin tumhare asli trained model ki file book ke sath nahi, isliye tumhara model aur dataset laga kar tumhe khud test karna hoga. class_indices.json ka format ( naam -> index ya index -> naam ) apni file me

print

kar ke dekh lo aur _classes wali line usi ke mutabiq rakho.

Karne ke kaam:​

  1. index.html me tasveer upload ka button/area aur preview banao, aur fetch("/predict/image", { method: "POST", body: formData }) se bhejo.

  2. Dono jawab (symptoms aur image) ek hi tarah ki result cards me dikhao.

  3. tests/ me image endpoint ka test likho (chhoti nakli tasveer bana kar) aur galat file type par 400 check karo.

  4. Imandari se likho: image model ki test accuracy README me likho. Agar wo kam hai (jildi bimariyon ke dataset aksar mushkil hote hain) to UI me saaf likho ke natija bharose ka nahi.

  5. Docker: TensorFlow ki wajah se image bohat bari ho jati hai. Dockerfile me tensorflow-cpu ya tensorflow ka fark likho aur image ka size compare karo (docker images).

40.10 Home-Based Project C: Monitoring aur Drift​

Maqsad: Chapter 39 ka kaam is project par laagu karo.

  1. /predict me jawab ke sath top prediction ki probability bhi log karo. Agar 0.5 se kam ho to WARNING likho ("kam bharosa").

  2. Ek global counter rakho jo kitni requests me kaun si bimari top par aayi ginta hai, aur GET /stats se dikhata hai. Kya koi ek bimari hamesha top par aati hai?

  3. Drift idea: training data me ek request me average symptoms kitni hoti hain (X.sum(axis=1).mean()) aur live requests me kitni? KS test (Chapter 39) se dono ki taqseem milao.

  4. Ek monitor_report.py likho jo app.log parh kar requests ki ginti, average latency aur WARNING ki ginti chhapta hai.

  5. Nijji maloomat (symptoms ki list) logs me na likho. Ye is task ka imtihan hai.

40.11 Home-Based Project D: Apna Model, Apni API​

Ab wohi poora raasta apne kisi purane project par chalao jo Module 4 me banaya tha (maslan Heart Disease, HR Attrition, Credit Card Churn, Titanic ). Aham baat: saari cheezein aik jaisi hon , sirf model aur input badlein.

QadamKaamChapter
1Model ko joblib se save karo (features kitarteebbhi)34
2FastAPI +Pydanticse /predict aur /health banao35
3Tests likho (TestClient + pytest)35
QadamKaamChapter
4Dockerfile, .dockerignore, requirements.txt (versions ke sath)36
5Training MLflow me log karo, achha model register karo37
6GitHub par push karo, ci.yml se hara tick lo, README par badge38
7Logging lagao, aurdrift ka ek checklikho39

Deliverable: GitHub repo (kam az kam: code, tests, Dockerfile, ci.yml, README), aur 1 screenshot hare CI ka . Agar hosting par chalana chahti ho (CD), to dhyan rakho ke free plans par memory kam hoti hai; bare model wahan nahi chalte.

Rubric (Home Project)Number
API aur tests (kam az kam 3 tests, galat input ka bhi)25
Docker image build hoti aur chalti hai20
MLflow me kam az kam 3 runs aur ek registered model20
CI hara hai aur README mekaise chalaensaaf likha hai20
Logging aur ek monitoring/drift check15
Kul100

40.12 Module 7 ka Mukhtasar​

ChapterMukhya IdeaAham Code / Tool
34: Introduction to MLOpsMLOps = ML + DevOps + Data Engineering; lifecycle ek
chakkar hai
joblib.dump / joblib.load,
requirements.txt
35: FastAPIModel ko API bana kar serve karna; input validation;
testing
FastAPI, BaseModel, Field, /docs,
TestClient
36: DockerApp aur environment ek dabbe me; image vs containerDockerfile, docker build, run, ps,
stop
37: MLflowExperiments ka hisaab aur model registrylog_param, log_metric, log_model,
mlflow ui, alias
38: CI/CDHar badlav par khud test aur build.github/workflows/ci.yml, pytest,
docker build
39: Logging & MonitoringBlack box (logs) aur live model ki sehat; driftlogging, AccuracyMonitor, KS test
(ks_2samp)
40: ProjectsSab kuch ek asli project me jodnaAI Medical Assistant (FastAPI + UI)

Agla qadam: Is module ke baad tum apne models ko deploy aur monitor karna jaanti ho. Agle module me Big Data aur Cloud ki taraf badhoge: jahan data itna bara hota hai ke ek computer me nahi samata, aur apps cloud par chalti hain.

40.13 Aam Galtiyan aur Hands-on​

AAM GALTIYAN​
  • Project ko "chal gaya" samajh kar tests, README aur disclaimer chhor dena.

  • Banawati data par aaye hue achhe numbers ko asli duniya ka daawa banana.

  • Medical jese masalon me disclaimer aur limitations na likhna.

  • Deploy kar ke monitoring na lagana.

  • Sirf top 1 jawab dikhana. Bohat sare classes me top 3 zyada imandar aur kaam ka hota hai.

Hands-on​

  • PROFILES me ek 10vi bimari jodo (maslan Chickenpox : fever, rash, itchy_eyes...), python train.py dobara chalao aur UI me test karo.

  • UI me "saaf karo (Clear)" button jodo jo sab chips hata de.

  • test_predict_returns_top3 jaisa ek aur test likho jo Allergy ki symptoms (sneezing, itchy_eyes) par Allergy ka top par aana check kare.

  • Container ko docker stop kar ke dobara run karo aur dekho ke logs (docker logs) me kya nazar aata hai.

  • README me "Limitations" ka hissa likho: kam az kam 4 baatein jo ye project nahi kar sakta.

KEY TAKEAWAYS​
  • AI Medical Assistant me ek hi project me FastAPI, UI, tests, Docker, CI aur logging jur gaye.

  • Class Project: model train, /predict, UI, 5 tests, Docker image, CI ka hara tick, aur saaf README.

  • Home Projects: A MLflow aur registry, B /predict/image, C monitoring/drift, D apna model apni API.

  • Medical jese masalon me disclaimer, limitations aur nijji data ki hifazat project ka hissa hain.

  • Banawati data par achhe numbers asli duniya ka saboot nahi; hamesha imandari se likho.