دليل هندسي: Freelance Python Programmer: High-Speed Analytics with Polars, DuckDB & Parquet. ألديان فازريهادي (Aldian Fazrihady) مبرمج مستقل أول ومهندس ذكاء اصطناعي (Python, Django, FastAPI, React, Next.js, TypeScript, Golang).

Replace slow OOM Pandas jobs by hiring senior freelance Python programmer Aldian Fazrihady to build zero-copy Polars, DuckDB, and Apache Parquet data pipelines.

import duckdb

def query_tenant_usage_metrics(parquet_glob_path: str, tenant_id: str) -> list[dict]:
    con = duckdb.connect(database=":memory:")
    query = """
        SELECT
            date_trunc('day', event_timestamp) AS usage_day,
            model_name,
            COUNT(*) AS total_requests,
            SUM(prompt_tokens + completion_tokens) AS total_tokens,
            QUANTILE_CONT(latency_ms, 0.95) AS p95_latency_ms
        FROM read_parquet($1, hive_partitioning = true)
        WHERE tenant_id = $2
        GROUP BY 1, 2
        ORDER BY 1 DESC, 4 DESC;
    """
    rel = con.execute(query, [parquet_glob_path, tenant_id])
    cols = [desc[0] for desc in rel.description]
    return [dict(zip(cols, row)) for row in rel.fetchall()]

توظيف عبر Upwork | دليل 450+ مقال هندسي