LeadMachine — Pipeline de prospection IA

1# Anti-detection scraping — human-like behavior simulation
2class StealthScraper:
3    def __init__(self):
4        self.proxy_pool = ProxyRotator(providers=["residential"])
5        self.browser = playwright.chromium.launch(
6            args=["--disable-blink-features=AutomationControlled"]
7        )
8
9    async def scrape_profile(self, url: str) -> CompanyData:
10        # Randomize viewport & user agent
11        context = await self.browser.new_context(
12            viewport=random.choice(VIEWPORTS),
13            user_agent=self.proxy_pool.get_ua(),
14        )
15        # Human-like delays (gaussian distribution)
16        await asyncio.sleep(random.gauss(mu=2.1, sigma=0.6))
17
18        page = await context.new_page()
19        await page.goto(url, wait_until="networkidle")
20
21        # Simulate scroll & mouse movement
22        await self._human_scroll(page)
23        return await self._extract_company_data(page)

1# Lead scoring — calibrated on client's historical deals
2import lightgbm as lgb
3from sklearn.preprocessing import LabelEncoder
4
5class LeadScorer:
6    FEATURES = [
7        "revenue_estimated", "headcount", "tech_stack_match",
8        "funding_signal", "hiring_signal", "industry_fit",
9        "decision_maker_detected", "days_since_founded",
10    ]
11
12    def __init__(self, model_path: str):
13        self.model = lgb.Booster(model_file=model_path)
14        self.enc   = LabelEncoder()
15
16    def score(self, lead: EnrichedLead) -> int:
17        features = [getattr(lead, f, 0) for f in self.FEATURES]
18        proba = self.model.predict([features])[0]
19        # Scale to 0–100, clamp
20        return min(100, max(0, int(proba * 100)))
21
22    def batch_score(self, leads: list[EnrichedLead]) -> list[ScoredLead]:
23        matrix = [[getattr(l, f, 0) for f in self.FEATURES] for l in leads]
24        probas = self.model.predict(matrix)
25        return [
26            ScoredLead(**lead.__dict__, score=int(p * 100))
27            for lead, p in zip(leads, probas)
28        ]